Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * NET An implementation of the SOCKET network access protocol. | |
3 | * | |
4 | * Version: @(#)socket.c 1.1.93 18/02/95 | |
5 | * | |
6 | * Authors: Orest Zborowski, <obz@Kodak.COM> | |
02c30a84 | 7 | * Ross Biro |
1da177e4 LT |
8 | * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> |
9 | * | |
10 | * Fixes: | |
11 | * Anonymous : NOTSOCK/BADF cleanup. Error fix in | |
12 | * shutdown() | |
13 | * Alan Cox : verify_area() fixes | |
14 | * Alan Cox : Removed DDI | |
15 | * Jonathan Kamens : SOCK_DGRAM reconnect bug | |
16 | * Alan Cox : Moved a load of checks to the very | |
17 | * top level. | |
18 | * Alan Cox : Move address structures to/from user | |
19 | * mode above the protocol layers. | |
20 | * Rob Janssen : Allow 0 length sends. | |
21 | * Alan Cox : Asynchronous I/O support (cribbed from the | |
22 | * tty drivers). | |
23 | * Niibe Yutaka : Asynchronous I/O for writes (4.4BSD style) | |
24 | * Jeff Uphoff : Made max number of sockets command-line | |
25 | * configurable. | |
26 | * Matti Aarnio : Made the number of sockets dynamic, | |
27 | * to be allocated when needed, and mr. | |
28 | * Uphoff's max is used as max to be | |
29 | * allowed to allocate. | |
30 | * Linus : Argh. removed all the socket allocation | |
31 | * altogether: it's in the inode now. | |
32 | * Alan Cox : Made sock_alloc()/sock_release() public | |
33 | * for NetROM and future kernel nfsd type | |
34 | * stuff. | |
35 | * Alan Cox : sendmsg/recvmsg basics. | |
36 | * Tom Dyas : Export net symbols. | |
37 | * Marcin Dalecki : Fixed problems with CONFIG_NET="n". | |
38 | * Alan Cox : Added thread locking to sys_* calls | |
39 | * for sockets. May have errors at the | |
40 | * moment. | |
41 | * Kevin Buhr : Fixed the dumb errors in the above. | |
42 | * Andi Kleen : Some small cleanups, optimizations, | |
43 | * and fixed a copy_from_user() bug. | |
44 | * Tigran Aivazian : sys_send(args) calls sys_sendto(args, NULL, 0) | |
89bddce5 | 45 | * Tigran Aivazian : Made listen(2) backlog sanity checks |
1da177e4 LT |
46 | * protocol-independent |
47 | * | |
48 | * | |
49 | * This program is free software; you can redistribute it and/or | |
50 | * modify it under the terms of the GNU General Public License | |
51 | * as published by the Free Software Foundation; either version | |
52 | * 2 of the License, or (at your option) any later version. | |
53 | * | |
54 | * | |
55 | * This module is effectively the top level interface to the BSD socket | |
89bddce5 | 56 | * paradigm. |
1da177e4 LT |
57 | * |
58 | * Based upon Swansea University Computer Society NET3.039 | |
59 | */ | |
60 | ||
1da177e4 | 61 | #include <linux/mm.h> |
1da177e4 LT |
62 | #include <linux/socket.h> |
63 | #include <linux/file.h> | |
64 | #include <linux/net.h> | |
65 | #include <linux/interrupt.h> | |
aaca0bdc | 66 | #include <linux/thread_info.h> |
55737fda | 67 | #include <linux/rcupdate.h> |
1da177e4 LT |
68 | #include <linux/netdevice.h> |
69 | #include <linux/proc_fs.h> | |
70 | #include <linux/seq_file.h> | |
4a3e2f71 | 71 | #include <linux/mutex.h> |
1da177e4 | 72 | #include <linux/if_bridge.h> |
20380731 ACM |
73 | #include <linux/if_frad.h> |
74 | #include <linux/if_vlan.h> | |
1da177e4 LT |
75 | #include <linux/init.h> |
76 | #include <linux/poll.h> | |
77 | #include <linux/cache.h> | |
78 | #include <linux/module.h> | |
79 | #include <linux/highmem.h> | |
1da177e4 LT |
80 | #include <linux/mount.h> |
81 | #include <linux/security.h> | |
82 | #include <linux/syscalls.h> | |
83 | #include <linux/compat.h> | |
84 | #include <linux/kmod.h> | |
3ec3b2fb | 85 | #include <linux/audit.h> |
d86b5e0e | 86 | #include <linux/wireless.h> |
1b8d7ae4 | 87 | #include <linux/nsproxy.h> |
1fd7317d | 88 | #include <linux/magic.h> |
5a0e3ad6 | 89 | #include <linux/slab.h> |
600e1779 | 90 | #include <linux/xattr.h> |
1da177e4 LT |
91 | |
92 | #include <asm/uaccess.h> | |
93 | #include <asm/unistd.h> | |
94 | ||
95 | #include <net/compat.h> | |
87de87d5 | 96 | #include <net/wext.h> |
f8451725 | 97 | #include <net/cls_cgroup.h> |
1da177e4 LT |
98 | |
99 | #include <net/sock.h> | |
100 | #include <linux/netfilter.h> | |
101 | ||
6b96018b AB |
102 | #include <linux/if_tun.h> |
103 | #include <linux/ipv6_route.h> | |
104 | #include <linux/route.h> | |
6b96018b AB |
105 | #include <linux/sockios.h> |
106 | #include <linux/atalk.h> | |
107 | ||
1da177e4 | 108 | static int sock_no_open(struct inode *irrelevant, struct file *dontcare); |
027445c3 BP |
109 | static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov, |
110 | unsigned long nr_segs, loff_t pos); | |
111 | static ssize_t sock_aio_write(struct kiocb *iocb, const struct iovec *iov, | |
112 | unsigned long nr_segs, loff_t pos); | |
89bddce5 | 113 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
114 | |
115 | static int sock_close(struct inode *inode, struct file *file); | |
116 | static unsigned int sock_poll(struct file *file, | |
117 | struct poll_table_struct *wait); | |
89bddce5 | 118 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
119 | #ifdef CONFIG_COMPAT |
120 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 121 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 122 | #endif |
1da177e4 | 123 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
124 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
125 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 126 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 127 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 128 | unsigned int flags); |
1da177e4 | 129 | |
1da177e4 LT |
130 | /* |
131 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
132 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
133 | */ | |
134 | ||
da7071d7 | 135 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
136 | .owner = THIS_MODULE, |
137 | .llseek = no_llseek, | |
138 | .aio_read = sock_aio_read, | |
139 | .aio_write = sock_aio_write, | |
140 | .poll = sock_poll, | |
141 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
142 | #ifdef CONFIG_COMPAT |
143 | .compat_ioctl = compat_sock_ioctl, | |
144 | #endif | |
1da177e4 LT |
145 | .mmap = sock_mmap, |
146 | .open = sock_no_open, /* special open code to disallow open via /proc */ | |
147 | .release = sock_close, | |
148 | .fasync = sock_fasync, | |
5274f052 JA |
149 | .sendpage = sock_sendpage, |
150 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 151 | .splice_read = sock_splice_read, |
1da177e4 LT |
152 | }; |
153 | ||
154 | /* | |
155 | * The protocol list. Each protocol is registered in here. | |
156 | */ | |
157 | ||
1da177e4 | 158 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 159 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 160 | |
1da177e4 LT |
161 | /* |
162 | * Statistics counters of the socket lists | |
163 | */ | |
164 | ||
c6d409cf | 165 | static DEFINE_PER_CPU(int, sockets_in_use); |
1da177e4 LT |
166 | |
167 | /* | |
89bddce5 SH |
168 | * Support routines. |
169 | * Move socket addresses back and forth across the kernel/user | |
170 | * divide and look after the messy bits. | |
1da177e4 LT |
171 | */ |
172 | ||
1da177e4 LT |
173 | /** |
174 | * move_addr_to_kernel - copy a socket address into kernel space | |
175 | * @uaddr: Address in user space | |
176 | * @kaddr: Address in kernel space | |
177 | * @ulen: Length in user space | |
178 | * | |
179 | * The address is copied into kernel space. If the provided address is | |
180 | * too long an error code of -EINVAL is returned. If the copy gives | |
181 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
182 | */ | |
183 | ||
43db362d | 184 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 185 | { |
230b1839 | 186 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 187 | return -EINVAL; |
89bddce5 | 188 | if (ulen == 0) |
1da177e4 | 189 | return 0; |
89bddce5 | 190 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 191 | return -EFAULT; |
3ec3b2fb | 192 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
193 | } |
194 | ||
195 | /** | |
196 | * move_addr_to_user - copy an address to user space | |
197 | * @kaddr: kernel space address | |
198 | * @klen: length of address in kernel | |
199 | * @uaddr: user space address | |
200 | * @ulen: pointer to user length field | |
201 | * | |
202 | * The value pointed to by ulen on entry is the buffer length available. | |
203 | * This is overwritten with the buffer space used. -EINVAL is returned | |
204 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
205 | * is returned if either the buffer or the length field are not | |
206 | * accessible. | |
207 | * After copying the data up to the limit the user specifies, the true | |
208 | * length of the data is written over the length limit the user | |
209 | * specified. Zero is returned for a success. | |
210 | */ | |
89bddce5 | 211 | |
43db362d | 212 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 213 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
214 | { |
215 | int err; | |
216 | int len; | |
217 | ||
5c586f16 | 218 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
219 | err = get_user(len, ulen); |
220 | if (err) | |
1da177e4 | 221 | return err; |
89bddce5 SH |
222 | if (len > klen) |
223 | len = klen; | |
5c586f16 | 224 | if (len < 0) |
1da177e4 | 225 | return -EINVAL; |
89bddce5 | 226 | if (len) { |
d6fe3945 SG |
227 | if (audit_sockaddr(klen, kaddr)) |
228 | return -ENOMEM; | |
89bddce5 | 229 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
230 | return -EFAULT; |
231 | } | |
232 | /* | |
89bddce5 SH |
233 | * "fromlen shall refer to the value before truncation.." |
234 | * 1003.1g | |
1da177e4 LT |
235 | */ |
236 | return __put_user(klen, ulen); | |
237 | } | |
238 | ||
e18b890b | 239 | static struct kmem_cache *sock_inode_cachep __read_mostly; |
1da177e4 LT |
240 | |
241 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
242 | { | |
243 | struct socket_alloc *ei; | |
eaefd110 | 244 | struct socket_wq *wq; |
89bddce5 | 245 | |
e94b1766 | 246 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
247 | if (!ei) |
248 | return NULL; | |
eaefd110 ED |
249 | wq = kmalloc(sizeof(*wq), GFP_KERNEL); |
250 | if (!wq) { | |
43815482 ED |
251 | kmem_cache_free(sock_inode_cachep, ei); |
252 | return NULL; | |
253 | } | |
eaefd110 ED |
254 | init_waitqueue_head(&wq->wait); |
255 | wq->fasync_list = NULL; | |
256 | RCU_INIT_POINTER(ei->socket.wq, wq); | |
89bddce5 | 257 | |
1da177e4 LT |
258 | ei->socket.state = SS_UNCONNECTED; |
259 | ei->socket.flags = 0; | |
260 | ei->socket.ops = NULL; | |
261 | ei->socket.sk = NULL; | |
262 | ei->socket.file = NULL; | |
1da177e4 LT |
263 | |
264 | return &ei->vfs_inode; | |
265 | } | |
266 | ||
267 | static void sock_destroy_inode(struct inode *inode) | |
268 | { | |
43815482 | 269 | struct socket_alloc *ei; |
eaefd110 | 270 | struct socket_wq *wq; |
43815482 ED |
271 | |
272 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
eaefd110 | 273 | wq = rcu_dereference_protected(ei->socket.wq, 1); |
61845220 | 274 | kfree_rcu(wq, rcu); |
43815482 | 275 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
276 | } |
277 | ||
51cc5068 | 278 | static void init_once(void *foo) |
1da177e4 | 279 | { |
89bddce5 | 280 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 281 | |
a35afb83 | 282 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 283 | } |
89bddce5 | 284 | |
1da177e4 LT |
285 | static int init_inodecache(void) |
286 | { | |
287 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
288 | sizeof(struct socket_alloc), |
289 | 0, | |
290 | (SLAB_HWCACHE_ALIGN | | |
291 | SLAB_RECLAIM_ACCOUNT | | |
292 | SLAB_MEM_SPREAD), | |
20c2df83 | 293 | init_once); |
1da177e4 LT |
294 | if (sock_inode_cachep == NULL) |
295 | return -ENOMEM; | |
296 | return 0; | |
297 | } | |
298 | ||
b87221de | 299 | static const struct super_operations sockfs_ops = { |
c6d409cf ED |
300 | .alloc_inode = sock_alloc_inode, |
301 | .destroy_inode = sock_destroy_inode, | |
302 | .statfs = simple_statfs, | |
1da177e4 LT |
303 | }; |
304 | ||
c23fbb6b ED |
305 | /* |
306 | * sockfs_dname() is called from d_path(). | |
307 | */ | |
308 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
309 | { | |
310 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
311 | dentry->d_inode->i_ino); | |
312 | } | |
313 | ||
3ba13d17 | 314 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 315 | .d_dname = sockfs_dname, |
1da177e4 LT |
316 | }; |
317 | ||
c74a1cbb AV |
318 | static struct dentry *sockfs_mount(struct file_system_type *fs_type, |
319 | int flags, const char *dev_name, void *data) | |
320 | { | |
321 | return mount_pseudo(fs_type, "socket:", &sockfs_ops, | |
322 | &sockfs_dentry_operations, SOCKFS_MAGIC); | |
323 | } | |
324 | ||
325 | static struct vfsmount *sock_mnt __read_mostly; | |
326 | ||
327 | static struct file_system_type sock_fs_type = { | |
328 | .name = "sockfs", | |
329 | .mount = sockfs_mount, | |
330 | .kill_sb = kill_anon_super, | |
331 | }; | |
332 | ||
1da177e4 LT |
333 | /* |
334 | * Obtains the first available file descriptor and sets it up for use. | |
335 | * | |
39d8c1b6 DM |
336 | * These functions create file structures and maps them to fd space |
337 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
338 | * and file struct implicitly stored in sock->file. |
339 | * Note that another thread may close file descriptor before we return | |
340 | * from this function. We use the fact that now we do not refer | |
341 | * to socket after mapping. If one day we will need it, this | |
342 | * function will increment ref. count on file by 1. | |
343 | * | |
344 | * In any case returned fd MAY BE not valid! | |
345 | * This race condition is unavoidable | |
346 | * with shared fd spaces, we cannot solve it inside kernel, | |
347 | * but we take care of internal coherence yet. | |
348 | */ | |
349 | ||
aab174f0 | 350 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 351 | { |
7cbe66b6 | 352 | struct qstr name = { .name = "" }; |
2c48b9c4 | 353 | struct path path; |
7cbe66b6 | 354 | struct file *file; |
1da177e4 | 355 | |
600e1779 MY |
356 | if (dname) { |
357 | name.name = dname; | |
358 | name.len = strlen(name.name); | |
359 | } else if (sock->sk) { | |
360 | name.name = sock->sk->sk_prot_creator->name; | |
361 | name.len = strlen(name.name); | |
362 | } | |
4b936885 | 363 | path.dentry = d_alloc_pseudo(sock_mnt->mnt_sb, &name); |
28407630 AV |
364 | if (unlikely(!path.dentry)) |
365 | return ERR_PTR(-ENOMEM); | |
2c48b9c4 | 366 | path.mnt = mntget(sock_mnt); |
39d8c1b6 | 367 | |
2c48b9c4 | 368 | d_instantiate(path.dentry, SOCK_INODE(sock)); |
cc3808f8 | 369 | SOCK_INODE(sock)->i_fop = &socket_file_ops; |
39d8c1b6 | 370 | |
2c48b9c4 | 371 | file = alloc_file(&path, FMODE_READ | FMODE_WRITE, |
ce8d2cdf | 372 | &socket_file_ops); |
39b65252 | 373 | if (unlikely(IS_ERR(file))) { |
cc3808f8 | 374 | /* drop dentry, keep inode */ |
7de9c6ee | 375 | ihold(path.dentry->d_inode); |
2c48b9c4 | 376 | path_put(&path); |
39b65252 | 377 | return file; |
cc3808f8 AV |
378 | } |
379 | ||
380 | sock->file = file; | |
77d27200 | 381 | file->f_flags = O_RDWR | (flags & O_NONBLOCK); |
39d8c1b6 | 382 | file->private_data = sock; |
28407630 | 383 | return file; |
39d8c1b6 | 384 | } |
56b31d1c | 385 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 386 | |
56b31d1c | 387 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
388 | { |
389 | struct file *newfile; | |
28407630 AV |
390 | int fd = get_unused_fd_flags(flags); |
391 | if (unlikely(fd < 0)) | |
392 | return fd; | |
39d8c1b6 | 393 | |
aab174f0 | 394 | newfile = sock_alloc_file(sock, flags, NULL); |
28407630 | 395 | if (likely(!IS_ERR(newfile))) { |
39d8c1b6 | 396 | fd_install(fd, newfile); |
28407630 AV |
397 | return fd; |
398 | } | |
7cbe66b6 | 399 | |
28407630 AV |
400 | put_unused_fd(fd); |
401 | return PTR_ERR(newfile); | |
1da177e4 LT |
402 | } |
403 | ||
406a3c63 | 404 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 405 | { |
6cb153ca BL |
406 | if (file->f_op == &socket_file_ops) |
407 | return file->private_data; /* set in sock_map_fd */ | |
408 | ||
23bb80d2 ED |
409 | *err = -ENOTSOCK; |
410 | return NULL; | |
6cb153ca | 411 | } |
406a3c63 | 412 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 413 | |
1da177e4 | 414 | /** |
c6d409cf | 415 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
416 | * @fd: file handle |
417 | * @err: pointer to an error code return | |
418 | * | |
419 | * The file handle passed in is locked and the socket it is bound | |
420 | * too is returned. If an error occurs the err pointer is overwritten | |
421 | * with a negative errno code and NULL is returned. The function checks | |
422 | * for both invalid handles and passing a handle which is not a socket. | |
423 | * | |
424 | * On a success the socket object pointer is returned. | |
425 | */ | |
426 | ||
427 | struct socket *sockfd_lookup(int fd, int *err) | |
428 | { | |
429 | struct file *file; | |
1da177e4 LT |
430 | struct socket *sock; |
431 | ||
89bddce5 SH |
432 | file = fget(fd); |
433 | if (!file) { | |
1da177e4 LT |
434 | *err = -EBADF; |
435 | return NULL; | |
436 | } | |
89bddce5 | 437 | |
6cb153ca BL |
438 | sock = sock_from_file(file, err); |
439 | if (!sock) | |
1da177e4 | 440 | fput(file); |
6cb153ca BL |
441 | return sock; |
442 | } | |
c6d409cf | 443 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 444 | |
6cb153ca BL |
445 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
446 | { | |
447 | struct file *file; | |
448 | struct socket *sock; | |
449 | ||
3672558c | 450 | *err = -EBADF; |
6cb153ca BL |
451 | file = fget_light(fd, fput_needed); |
452 | if (file) { | |
453 | sock = sock_from_file(file, err); | |
454 | if (sock) | |
455 | return sock; | |
456 | fput_light(file, *fput_needed); | |
1da177e4 | 457 | } |
6cb153ca | 458 | return NULL; |
1da177e4 LT |
459 | } |
460 | ||
600e1779 MY |
461 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" |
462 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
463 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
464 | static ssize_t sockfs_getxattr(struct dentry *dentry, | |
465 | const char *name, void *value, size_t size) | |
466 | { | |
467 | const char *proto_name; | |
468 | size_t proto_size; | |
469 | int error; | |
470 | ||
471 | error = -ENODATA; | |
472 | if (!strncmp(name, XATTR_NAME_SOCKPROTONAME, XATTR_NAME_SOCKPROTONAME_LEN)) { | |
473 | proto_name = dentry->d_name.name; | |
474 | proto_size = strlen(proto_name); | |
475 | ||
476 | if (value) { | |
477 | error = -ERANGE; | |
478 | if (proto_size + 1 > size) | |
479 | goto out; | |
480 | ||
481 | strncpy(value, proto_name, proto_size + 1); | |
482 | } | |
483 | error = proto_size + 1; | |
484 | } | |
485 | ||
486 | out: | |
487 | return error; | |
488 | } | |
489 | ||
490 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, | |
491 | size_t size) | |
492 | { | |
493 | ssize_t len; | |
494 | ssize_t used = 0; | |
495 | ||
496 | len = security_inode_listsecurity(dentry->d_inode, buffer, size); | |
497 | if (len < 0) | |
498 | return len; | |
499 | used += len; | |
500 | if (buffer) { | |
501 | if (size < used) | |
502 | return -ERANGE; | |
503 | buffer += len; | |
504 | } | |
505 | ||
506 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
507 | used += len; | |
508 | if (buffer) { | |
509 | if (size < used) | |
510 | return -ERANGE; | |
511 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
512 | buffer += len; | |
513 | } | |
514 | ||
515 | return used; | |
516 | } | |
517 | ||
c9beb246 LC |
518 | int sockfs_setattr(struct dentry *dentry, struct iattr *iattr) |
519 | { | |
520 | int err = simple_setattr(dentry, iattr); | |
521 | ||
522 | if (!err) { | |
523 | struct socket *sock = SOCKET_I(dentry->d_inode); | |
524 | ||
525 | sock->sk->sk_uid = iattr->ia_uid; | |
526 | } | |
527 | ||
528 | return err; | |
529 | } | |
530 | ||
600e1779 MY |
531 | static const struct inode_operations sockfs_inode_ops = { |
532 | .getxattr = sockfs_getxattr, | |
533 | .listxattr = sockfs_listxattr, | |
c9beb246 | 534 | .setattr = sockfs_setattr, |
600e1779 MY |
535 | }; |
536 | ||
1da177e4 LT |
537 | /** |
538 | * sock_alloc - allocate a socket | |
89bddce5 | 539 | * |
1da177e4 LT |
540 | * Allocate a new inode and socket object. The two are bound together |
541 | * and initialised. The socket is then returned. If we are out of inodes | |
542 | * NULL is returned. | |
543 | */ | |
544 | ||
545 | static struct socket *sock_alloc(void) | |
546 | { | |
89bddce5 SH |
547 | struct inode *inode; |
548 | struct socket *sock; | |
1da177e4 | 549 | |
a209dfc7 | 550 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
551 | if (!inode) |
552 | return NULL; | |
553 | ||
554 | sock = SOCKET_I(inode); | |
555 | ||
29a020d3 | 556 | kmemcheck_annotate_bitfield(sock, type); |
85fe4025 | 557 | inode->i_ino = get_next_ino(); |
89bddce5 | 558 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
559 | inode->i_uid = current_fsuid(); |
560 | inode->i_gid = current_fsgid(); | |
600e1779 | 561 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 562 | |
19e8d69c | 563 | this_cpu_add(sockets_in_use, 1); |
1da177e4 LT |
564 | return sock; |
565 | } | |
566 | ||
567 | /* | |
568 | * In theory you can't get an open on this inode, but /proc provides | |
569 | * a back door. Remember to keep it shut otherwise you'll let the | |
570 | * creepy crawlies in. | |
571 | */ | |
89bddce5 | 572 | |
1da177e4 LT |
573 | static int sock_no_open(struct inode *irrelevant, struct file *dontcare) |
574 | { | |
575 | return -ENXIO; | |
576 | } | |
577 | ||
4b6f5d20 | 578 | const struct file_operations bad_sock_fops = { |
1da177e4 LT |
579 | .owner = THIS_MODULE, |
580 | .open = sock_no_open, | |
6038f373 | 581 | .llseek = noop_llseek, |
1da177e4 LT |
582 | }; |
583 | ||
584 | /** | |
585 | * sock_release - close a socket | |
586 | * @sock: socket to close | |
587 | * | |
588 | * The socket is released from the protocol stack if it has a release | |
589 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 590 | * an inode not a file. |
1da177e4 | 591 | */ |
89bddce5 | 592 | |
1da177e4 LT |
593 | void sock_release(struct socket *sock) |
594 | { | |
595 | if (sock->ops) { | |
596 | struct module *owner = sock->ops->owner; | |
597 | ||
598 | sock->ops->release(sock); | |
599 | sock->ops = NULL; | |
600 | module_put(owner); | |
601 | } | |
602 | ||
eaefd110 | 603 | if (rcu_dereference_protected(sock->wq, 1)->fasync_list) |
1da177e4 LT |
604 | printk(KERN_ERR "sock_release: fasync list not empty!\n"); |
605 | ||
b09e786b MP |
606 | if (test_bit(SOCK_EXTERNALLY_ALLOCATED, &sock->flags)) |
607 | return; | |
608 | ||
19e8d69c | 609 | this_cpu_sub(sockets_in_use, 1); |
1da177e4 LT |
610 | if (!sock->file) { |
611 | iput(SOCK_INODE(sock)); | |
612 | return; | |
613 | } | |
89bddce5 | 614 | sock->file = NULL; |
1da177e4 | 615 | } |
c6d409cf | 616 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 617 | |
bf84a010 | 618 | void sock_tx_timestamp(struct sock *sk, __u8 *tx_flags) |
20d49473 | 619 | { |
2244d07b | 620 | *tx_flags = 0; |
20d49473 | 621 | if (sock_flag(sk, SOCK_TIMESTAMPING_TX_HARDWARE)) |
2244d07b | 622 | *tx_flags |= SKBTX_HW_TSTAMP; |
20d49473 | 623 | if (sock_flag(sk, SOCK_TIMESTAMPING_TX_SOFTWARE)) |
2244d07b | 624 | *tx_flags |= SKBTX_SW_TSTAMP; |
6e3e939f JB |
625 | if (sock_flag(sk, SOCK_WIFI_STATUS)) |
626 | *tx_flags |= SKBTX_WIFI_STATUS; | |
20d49473 PO |
627 | } |
628 | EXPORT_SYMBOL(sock_tx_timestamp); | |
629 | ||
228e548e AB |
630 | static inline int __sock_sendmsg_nosec(struct kiocb *iocb, struct socket *sock, |
631 | struct msghdr *msg, size_t size) | |
1da177e4 LT |
632 | { |
633 | struct sock_iocb *si = kiocb_to_siocb(iocb); | |
1da177e4 LT |
634 | |
635 | si->sock = sock; | |
636 | si->scm = NULL; | |
637 | si->msg = msg; | |
638 | si->size = size; | |
639 | ||
1da177e4 LT |
640 | return sock->ops->sendmsg(iocb, sock, msg, size); |
641 | } | |
642 | ||
228e548e AB |
643 | static inline int __sock_sendmsg(struct kiocb *iocb, struct socket *sock, |
644 | struct msghdr *msg, size_t size) | |
645 | { | |
646 | int err = security_socket_sendmsg(sock, msg, size); | |
647 | ||
648 | return err ?: __sock_sendmsg_nosec(iocb, sock, msg, size); | |
649 | } | |
650 | ||
1da177e4 LT |
651 | int sock_sendmsg(struct socket *sock, struct msghdr *msg, size_t size) |
652 | { | |
653 | struct kiocb iocb; | |
654 | struct sock_iocb siocb; | |
655 | int ret; | |
656 | ||
657 | init_sync_kiocb(&iocb, NULL); | |
658 | iocb.private = &siocb; | |
659 | ret = __sock_sendmsg(&iocb, sock, msg, size); | |
660 | if (-EIOCBQUEUED == ret) | |
661 | ret = wait_on_sync_kiocb(&iocb); | |
662 | return ret; | |
663 | } | |
c6d409cf | 664 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 | 665 | |
894dc24c | 666 | static int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg, size_t size) |
228e548e AB |
667 | { |
668 | struct kiocb iocb; | |
669 | struct sock_iocb siocb; | |
670 | int ret; | |
671 | ||
672 | init_sync_kiocb(&iocb, NULL); | |
673 | iocb.private = &siocb; | |
674 | ret = __sock_sendmsg_nosec(&iocb, sock, msg, size); | |
675 | if (-EIOCBQUEUED == ret) | |
676 | ret = wait_on_sync_kiocb(&iocb); | |
677 | return ret; | |
678 | } | |
679 | ||
1da177e4 LT |
680 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
681 | struct kvec *vec, size_t num, size_t size) | |
682 | { | |
683 | mm_segment_t oldfs = get_fs(); | |
684 | int result; | |
685 | ||
686 | set_fs(KERNEL_DS); | |
687 | /* | |
688 | * the following is safe, since for compiler definitions of kvec and | |
689 | * iovec are identical, yielding the same in-core layout and alignment | |
690 | */ | |
89bddce5 | 691 | msg->msg_iov = (struct iovec *)vec; |
1da177e4 LT |
692 | msg->msg_iovlen = num; |
693 | result = sock_sendmsg(sock, msg, size); | |
694 | set_fs(oldfs); | |
695 | return result; | |
696 | } | |
c6d409cf | 697 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 698 | |
92f37fd2 ED |
699 | /* |
700 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
701 | */ | |
702 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
703 | struct sk_buff *skb) | |
704 | { | |
20d49473 PO |
705 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
706 | struct timespec ts[3]; | |
707 | int empty = 1; | |
708 | struct skb_shared_hwtstamps *shhwtstamps = | |
709 | skb_hwtstamps(skb); | |
710 | ||
711 | /* Race occurred between timestamp enabling and packet | |
712 | receiving. Fill in the current time for now. */ | |
713 | if (need_software_tstamp && skb->tstamp.tv64 == 0) | |
714 | __net_timestamp(skb); | |
715 | ||
716 | if (need_software_tstamp) { | |
717 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
718 | struct timeval tv; | |
719 | skb_get_timestamp(skb, &tv); | |
720 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMP, | |
721 | sizeof(tv), &tv); | |
722 | } else { | |
842509b8 | 723 | skb_get_timestampns(skb, &ts[0]); |
20d49473 | 724 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPNS, |
842509b8 | 725 | sizeof(ts[0]), &ts[0]); |
20d49473 PO |
726 | } |
727 | } | |
728 | ||
729 | ||
730 | memset(ts, 0, sizeof(ts)); | |
6e94d1ef DB |
731 | if (sock_flag(sk, SOCK_TIMESTAMPING_SOFTWARE) && |
732 | ktime_to_timespec_cond(skb->tstamp, ts + 0)) | |
20d49473 | 733 | empty = 0; |
20d49473 PO |
734 | if (shhwtstamps) { |
735 | if (sock_flag(sk, SOCK_TIMESTAMPING_SYS_HARDWARE) && | |
6e94d1ef | 736 | ktime_to_timespec_cond(shhwtstamps->syststamp, ts + 1)) |
20d49473 PO |
737 | empty = 0; |
738 | if (sock_flag(sk, SOCK_TIMESTAMPING_RAW_HARDWARE) && | |
6e94d1ef | 739 | ktime_to_timespec_cond(shhwtstamps->hwtstamp, ts + 2)) |
20d49473 | 740 | empty = 0; |
92f37fd2 | 741 | } |
20d49473 PO |
742 | if (!empty) |
743 | put_cmsg(msg, SOL_SOCKET, | |
744 | SCM_TIMESTAMPING, sizeof(ts), &ts); | |
92f37fd2 | 745 | } |
7c81fd8b ACM |
746 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
747 | ||
6e3e939f JB |
748 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
749 | struct sk_buff *skb) | |
750 | { | |
751 | int ack; | |
752 | ||
753 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
754 | return; | |
755 | if (!skb->wifi_acked_valid) | |
756 | return; | |
757 | ||
758 | ack = skb->wifi_acked; | |
759 | ||
760 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
761 | } | |
762 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
763 | ||
11165f14 | 764 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
765 | struct sk_buff *skb) | |
3b885787 NH |
766 | { |
767 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && skb->dropcount) | |
768 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, | |
769 | sizeof(__u32), &skb->dropcount); | |
770 | } | |
771 | ||
767dd033 | 772 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
773 | struct sk_buff *skb) |
774 | { | |
775 | sock_recv_timestamp(msg, sk, skb); | |
776 | sock_recv_drops(msg, sk, skb); | |
777 | } | |
767dd033 | 778 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 779 | |
a2e27255 ACM |
780 | static inline int __sock_recvmsg_nosec(struct kiocb *iocb, struct socket *sock, |
781 | struct msghdr *msg, size_t size, int flags) | |
1da177e4 | 782 | { |
1da177e4 LT |
783 | struct sock_iocb *si = kiocb_to_siocb(iocb); |
784 | ||
785 | si->sock = sock; | |
786 | si->scm = NULL; | |
787 | si->msg = msg; | |
788 | si->size = size; | |
789 | si->flags = flags; | |
790 | ||
1da177e4 LT |
791 | return sock->ops->recvmsg(iocb, sock, msg, size, flags); |
792 | } | |
793 | ||
a2e27255 ACM |
794 | static inline int __sock_recvmsg(struct kiocb *iocb, struct socket *sock, |
795 | struct msghdr *msg, size_t size, int flags) | |
796 | { | |
797 | int err = security_socket_recvmsg(sock, msg, size, flags); | |
798 | ||
799 | return err ?: __sock_recvmsg_nosec(iocb, sock, msg, size, flags); | |
800 | } | |
801 | ||
89bddce5 | 802 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, |
1da177e4 LT |
803 | size_t size, int flags) |
804 | { | |
805 | struct kiocb iocb; | |
806 | struct sock_iocb siocb; | |
807 | int ret; | |
808 | ||
89bddce5 | 809 | init_sync_kiocb(&iocb, NULL); |
1da177e4 LT |
810 | iocb.private = &siocb; |
811 | ret = __sock_recvmsg(&iocb, sock, msg, size, flags); | |
812 | if (-EIOCBQUEUED == ret) | |
813 | ret = wait_on_sync_kiocb(&iocb); | |
814 | return ret; | |
815 | } | |
c6d409cf | 816 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 817 | |
a2e27255 ACM |
818 | static int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
819 | size_t size, int flags) | |
820 | { | |
821 | struct kiocb iocb; | |
822 | struct sock_iocb siocb; | |
823 | int ret; | |
824 | ||
825 | init_sync_kiocb(&iocb, NULL); | |
826 | iocb.private = &siocb; | |
827 | ret = __sock_recvmsg_nosec(&iocb, sock, msg, size, flags); | |
828 | if (-EIOCBQUEUED == ret) | |
829 | ret = wait_on_sync_kiocb(&iocb); | |
830 | return ret; | |
831 | } | |
832 | ||
c1249c0a ML |
833 | /** |
834 | * kernel_recvmsg - Receive a message from a socket (kernel space) | |
835 | * @sock: The socket to receive the message from | |
836 | * @msg: Received message | |
837 | * @vec: Input s/g array for message data | |
838 | * @num: Size of input s/g array | |
839 | * @size: Number of bytes to read | |
840 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
841 | * | |
842 | * On return the msg structure contains the scatter/gather array passed in the | |
843 | * vec argument. The array is modified so that it consists of the unfilled | |
844 | * portion of the original array. | |
845 | * | |
846 | * The returned value is the total number of bytes received, or an error. | |
847 | */ | |
89bddce5 SH |
848 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
849 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
850 | { |
851 | mm_segment_t oldfs = get_fs(); | |
852 | int result; | |
853 | ||
854 | set_fs(KERNEL_DS); | |
855 | /* | |
856 | * the following is safe, since for compiler definitions of kvec and | |
857 | * iovec are identical, yielding the same in-core layout and alignment | |
858 | */ | |
89bddce5 | 859 | msg->msg_iov = (struct iovec *)vec, msg->msg_iovlen = num; |
1da177e4 LT |
860 | result = sock_recvmsg(sock, msg, size, flags); |
861 | set_fs(oldfs); | |
862 | return result; | |
863 | } | |
c6d409cf | 864 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 LT |
865 | |
866 | static void sock_aio_dtor(struct kiocb *iocb) | |
867 | { | |
868 | kfree(iocb->private); | |
869 | } | |
870 | ||
ce1d4d3e CH |
871 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
872 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 873 | { |
1da177e4 LT |
874 | struct socket *sock; |
875 | int flags; | |
876 | ||
ce1d4d3e CH |
877 | sock = file->private_data; |
878 | ||
35f9c09f ED |
879 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
880 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
881 | flags |= more; | |
ce1d4d3e | 882 | |
e6949583 | 883 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 884 | } |
1da177e4 | 885 | |
9c55e01c | 886 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 887 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
888 | unsigned int flags) |
889 | { | |
890 | struct socket *sock = file->private_data; | |
891 | ||
997b37da RDC |
892 | if (unlikely(!sock->ops->splice_read)) |
893 | return -EINVAL; | |
894 | ||
9c55e01c JA |
895 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
896 | } | |
897 | ||
ce1d4d3e | 898 | static struct sock_iocb *alloc_sock_iocb(struct kiocb *iocb, |
89bddce5 | 899 | struct sock_iocb *siocb) |
ce1d4d3e CH |
900 | { |
901 | if (!is_sync_kiocb(iocb)) { | |
902 | siocb = kmalloc(sizeof(*siocb), GFP_KERNEL); | |
903 | if (!siocb) | |
904 | return NULL; | |
1da177e4 LT |
905 | iocb->ki_dtor = sock_aio_dtor; |
906 | } | |
1da177e4 | 907 | |
ce1d4d3e | 908 | siocb->kiocb = iocb; |
ce1d4d3e CH |
909 | iocb->private = siocb; |
910 | return siocb; | |
1da177e4 LT |
911 | } |
912 | ||
ce1d4d3e | 913 | static ssize_t do_sock_read(struct msghdr *msg, struct kiocb *iocb, |
027445c3 BP |
914 | struct file *file, const struct iovec *iov, |
915 | unsigned long nr_segs) | |
ce1d4d3e CH |
916 | { |
917 | struct socket *sock = file->private_data; | |
918 | size_t size = 0; | |
919 | int i; | |
1da177e4 | 920 | |
89bddce5 SH |
921 | for (i = 0; i < nr_segs; i++) |
922 | size += iov[i].iov_len; | |
1da177e4 | 923 | |
ce1d4d3e CH |
924 | msg->msg_name = NULL; |
925 | msg->msg_namelen = 0; | |
926 | msg->msg_control = NULL; | |
927 | msg->msg_controllen = 0; | |
89bddce5 | 928 | msg->msg_iov = (struct iovec *)iov; |
ce1d4d3e CH |
929 | msg->msg_iovlen = nr_segs; |
930 | msg->msg_flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; | |
931 | ||
932 | return __sock_recvmsg(iocb, sock, msg, size, msg->msg_flags); | |
933 | } | |
934 | ||
027445c3 BP |
935 | static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov, |
936 | unsigned long nr_segs, loff_t pos) | |
ce1d4d3e CH |
937 | { |
938 | struct sock_iocb siocb, *x; | |
939 | ||
1da177e4 LT |
940 | if (pos != 0) |
941 | return -ESPIPE; | |
027445c3 BP |
942 | |
943 | if (iocb->ki_left == 0) /* Match SYS5 behaviour */ | |
1da177e4 LT |
944 | return 0; |
945 | ||
027445c3 BP |
946 | |
947 | x = alloc_sock_iocb(iocb, &siocb); | |
ce1d4d3e CH |
948 | if (!x) |
949 | return -ENOMEM; | |
027445c3 | 950 | return do_sock_read(&x->async_msg, iocb, iocb->ki_filp, iov, nr_segs); |
1da177e4 LT |
951 | } |
952 | ||
ce1d4d3e | 953 | static ssize_t do_sock_write(struct msghdr *msg, struct kiocb *iocb, |
027445c3 BP |
954 | struct file *file, const struct iovec *iov, |
955 | unsigned long nr_segs) | |
1da177e4 | 956 | { |
ce1d4d3e CH |
957 | struct socket *sock = file->private_data; |
958 | size_t size = 0; | |
959 | int i; | |
1da177e4 | 960 | |
89bddce5 SH |
961 | for (i = 0; i < nr_segs; i++) |
962 | size += iov[i].iov_len; | |
1da177e4 | 963 | |
ce1d4d3e CH |
964 | msg->msg_name = NULL; |
965 | msg->msg_namelen = 0; | |
966 | msg->msg_control = NULL; | |
967 | msg->msg_controllen = 0; | |
89bddce5 | 968 | msg->msg_iov = (struct iovec *)iov; |
ce1d4d3e CH |
969 | msg->msg_iovlen = nr_segs; |
970 | msg->msg_flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; | |
971 | if (sock->type == SOCK_SEQPACKET) | |
972 | msg->msg_flags |= MSG_EOR; | |
1da177e4 | 973 | |
ce1d4d3e | 974 | return __sock_sendmsg(iocb, sock, msg, size); |
1da177e4 LT |
975 | } |
976 | ||
027445c3 BP |
977 | static ssize_t sock_aio_write(struct kiocb *iocb, const struct iovec *iov, |
978 | unsigned long nr_segs, loff_t pos) | |
ce1d4d3e CH |
979 | { |
980 | struct sock_iocb siocb, *x; | |
1da177e4 | 981 | |
ce1d4d3e CH |
982 | if (pos != 0) |
983 | return -ESPIPE; | |
027445c3 | 984 | |
027445c3 | 985 | x = alloc_sock_iocb(iocb, &siocb); |
ce1d4d3e CH |
986 | if (!x) |
987 | return -ENOMEM; | |
1da177e4 | 988 | |
027445c3 | 989 | return do_sock_write(&x->async_msg, iocb, iocb->ki_filp, iov, nr_segs); |
1da177e4 LT |
990 | } |
991 | ||
1da177e4 LT |
992 | /* |
993 | * Atomic setting of ioctl hooks to avoid race | |
994 | * with module unload. | |
995 | */ | |
996 | ||
4a3e2f71 | 997 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 998 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 999 | |
881d966b | 1000 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 1001 | { |
4a3e2f71 | 1002 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 1003 | br_ioctl_hook = hook; |
4a3e2f71 | 1004 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1005 | } |
1006 | EXPORT_SYMBOL(brioctl_set); | |
1007 | ||
4a3e2f71 | 1008 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 1009 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 1010 | |
881d966b | 1011 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 1012 | { |
4a3e2f71 | 1013 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1014 | vlan_ioctl_hook = hook; |
4a3e2f71 | 1015 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
1016 | } |
1017 | EXPORT_SYMBOL(vlan_ioctl_set); | |
1018 | ||
4a3e2f71 | 1019 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 1020 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 1021 | |
89bddce5 | 1022 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 1023 | { |
4a3e2f71 | 1024 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 1025 | dlci_ioctl_hook = hook; |
4a3e2f71 | 1026 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
1027 | } |
1028 | EXPORT_SYMBOL(dlci_ioctl_set); | |
1029 | ||
6b96018b AB |
1030 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
1031 | unsigned int cmd, unsigned long arg) | |
1032 | { | |
1033 | int err; | |
1034 | void __user *argp = (void __user *)arg; | |
1035 | ||
1036 | err = sock->ops->ioctl(sock, cmd, arg); | |
1037 | ||
1038 | /* | |
1039 | * If this ioctl is unknown try to hand it down | |
1040 | * to the NIC driver. | |
1041 | */ | |
1042 | if (err == -ENOIOCTLCMD) | |
1043 | err = dev_ioctl(net, cmd, argp); | |
1044 | ||
1045 | return err; | |
1046 | } | |
1047 | ||
1da177e4 LT |
1048 | /* |
1049 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1050 | * what to do with it - that's up to the protocol still. | |
1051 | */ | |
1052 | ||
1053 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) | |
1054 | { | |
1055 | struct socket *sock; | |
881d966b | 1056 | struct sock *sk; |
1da177e4 LT |
1057 | void __user *argp = (void __user *)arg; |
1058 | int pid, err; | |
881d966b | 1059 | struct net *net; |
1da177e4 | 1060 | |
b69aee04 | 1061 | sock = file->private_data; |
881d966b | 1062 | sk = sock->sk; |
3b1e0a65 | 1063 | net = sock_net(sk); |
1da177e4 | 1064 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) { |
881d966b | 1065 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 1066 | } else |
3d23e349 | 1067 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1068 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
881d966b | 1069 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 1070 | } else |
3d23e349 | 1071 | #endif |
89bddce5 | 1072 | switch (cmd) { |
1da177e4 LT |
1073 | case FIOSETOWN: |
1074 | case SIOCSPGRP: | |
1075 | err = -EFAULT; | |
1076 | if (get_user(pid, (int __user *)argp)) | |
1077 | break; | |
1078 | err = f_setown(sock->file, pid, 1); | |
1079 | break; | |
1080 | case FIOGETOWN: | |
1081 | case SIOCGPGRP: | |
609d7fa9 | 1082 | err = put_user(f_getown(sock->file), |
89bddce5 | 1083 | (int __user *)argp); |
1da177e4 LT |
1084 | break; |
1085 | case SIOCGIFBR: | |
1086 | case SIOCSIFBR: | |
1087 | case SIOCBRADDBR: | |
1088 | case SIOCBRDELBR: | |
1089 | err = -ENOPKG; | |
1090 | if (!br_ioctl_hook) | |
1091 | request_module("bridge"); | |
1092 | ||
4a3e2f71 | 1093 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1094 | if (br_ioctl_hook) |
881d966b | 1095 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1096 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1097 | break; |
1098 | case SIOCGIFVLAN: | |
1099 | case SIOCSIFVLAN: | |
1100 | err = -ENOPKG; | |
1101 | if (!vlan_ioctl_hook) | |
1102 | request_module("8021q"); | |
1103 | ||
4a3e2f71 | 1104 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1105 | if (vlan_ioctl_hook) |
881d966b | 1106 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1107 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1108 | break; |
1da177e4 LT |
1109 | case SIOCADDDLCI: |
1110 | case SIOCDELDLCI: | |
1111 | err = -ENOPKG; | |
1112 | if (!dlci_ioctl_hook) | |
1113 | request_module("dlci"); | |
1114 | ||
7512cbf6 PE |
1115 | mutex_lock(&dlci_ioctl_mutex); |
1116 | if (dlci_ioctl_hook) | |
1da177e4 | 1117 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1118 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
1119 | break; |
1120 | default: | |
6b96018b | 1121 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1122 | break; |
89bddce5 | 1123 | } |
1da177e4 LT |
1124 | return err; |
1125 | } | |
1126 | ||
1127 | int sock_create_lite(int family, int type, int protocol, struct socket **res) | |
1128 | { | |
1129 | int err; | |
1130 | struct socket *sock = NULL; | |
89bddce5 | 1131 | |
1da177e4 LT |
1132 | err = security_socket_create(family, type, protocol, 1); |
1133 | if (err) | |
1134 | goto out; | |
1135 | ||
1136 | sock = sock_alloc(); | |
1137 | if (!sock) { | |
1138 | err = -ENOMEM; | |
1139 | goto out; | |
1140 | } | |
1141 | ||
1da177e4 | 1142 | sock->type = type; |
7420ed23 VY |
1143 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1144 | if (err) | |
1145 | goto out_release; | |
1146 | ||
1da177e4 LT |
1147 | out: |
1148 | *res = sock; | |
1149 | return err; | |
7420ed23 VY |
1150 | out_release: |
1151 | sock_release(sock); | |
1152 | sock = NULL; | |
1153 | goto out; | |
1da177e4 | 1154 | } |
c6d409cf | 1155 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1156 | |
1157 | /* No kernel lock held - perfect */ | |
89bddce5 | 1158 | static unsigned int sock_poll(struct file *file, poll_table *wait) |
1da177e4 LT |
1159 | { |
1160 | struct socket *sock; | |
1161 | ||
1162 | /* | |
89bddce5 | 1163 | * We can't return errors to poll, so it's either yes or no. |
1da177e4 | 1164 | */ |
b69aee04 | 1165 | sock = file->private_data; |
1da177e4 LT |
1166 | return sock->ops->poll(file, sock, wait); |
1167 | } | |
1168 | ||
89bddce5 | 1169 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1170 | { |
b69aee04 | 1171 | struct socket *sock = file->private_data; |
1da177e4 LT |
1172 | |
1173 | return sock->ops->mmap(file, sock, vma); | |
1174 | } | |
1175 | ||
20380731 | 1176 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1177 | { |
1da177e4 LT |
1178 | sock_release(SOCKET_I(inode)); |
1179 | return 0; | |
1180 | } | |
1181 | ||
1182 | /* | |
1183 | * Update the socket async list | |
1184 | * | |
1185 | * Fasync_list locking strategy. | |
1186 | * | |
1187 | * 1. fasync_list is modified only under process context socket lock | |
1188 | * i.e. under semaphore. | |
1189 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1190 | * or under socket lock |
1da177e4 LT |
1191 | */ |
1192 | ||
1193 | static int sock_fasync(int fd, struct file *filp, int on) | |
1194 | { | |
989a2979 ED |
1195 | struct socket *sock = filp->private_data; |
1196 | struct sock *sk = sock->sk; | |
eaefd110 | 1197 | struct socket_wq *wq; |
1da177e4 | 1198 | |
989a2979 | 1199 | if (sk == NULL) |
1da177e4 | 1200 | return -EINVAL; |
1da177e4 LT |
1201 | |
1202 | lock_sock(sk); | |
eaefd110 ED |
1203 | wq = rcu_dereference_protected(sock->wq, sock_owned_by_user(sk)); |
1204 | fasync_helper(fd, filp, on, &wq->fasync_list); | |
1da177e4 | 1205 | |
eaefd110 | 1206 | if (!wq->fasync_list) |
989a2979 ED |
1207 | sock_reset_flag(sk, SOCK_FASYNC); |
1208 | else | |
bcdce719 | 1209 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1210 | |
989a2979 | 1211 | release_sock(sk); |
1da177e4 LT |
1212 | return 0; |
1213 | } | |
1214 | ||
43815482 | 1215 | /* This function may be called only under socket lock or callback_lock or rcu_lock */ |
1da177e4 LT |
1216 | |
1217 | int sock_wake_async(struct socket *sock, int how, int band) | |
1218 | { | |
43815482 ED |
1219 | struct socket_wq *wq; |
1220 | ||
1221 | if (!sock) | |
1222 | return -1; | |
1223 | rcu_read_lock(); | |
1224 | wq = rcu_dereference(sock->wq); | |
1225 | if (!wq || !wq->fasync_list) { | |
1226 | rcu_read_unlock(); | |
1da177e4 | 1227 | return -1; |
43815482 | 1228 | } |
89bddce5 | 1229 | switch (how) { |
8d8ad9d7 | 1230 | case SOCK_WAKE_WAITD: |
1da177e4 LT |
1231 | if (test_bit(SOCK_ASYNC_WAITDATA, &sock->flags)) |
1232 | break; | |
1233 | goto call_kill; | |
8d8ad9d7 | 1234 | case SOCK_WAKE_SPACE: |
1da177e4 LT |
1235 | if (!test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags)) |
1236 | break; | |
1237 | /* fall through */ | |
8d8ad9d7 | 1238 | case SOCK_WAKE_IO: |
89bddce5 | 1239 | call_kill: |
43815482 | 1240 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1241 | break; |
8d8ad9d7 | 1242 | case SOCK_WAKE_URG: |
43815482 | 1243 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1244 | } |
43815482 | 1245 | rcu_read_unlock(); |
1da177e4 LT |
1246 | return 0; |
1247 | } | |
c6d409cf | 1248 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1249 | |
721db93a | 1250 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1251 | struct socket **res, int kern) |
1da177e4 LT |
1252 | { |
1253 | int err; | |
1254 | struct socket *sock; | |
55737fda | 1255 | const struct net_proto_family *pf; |
1da177e4 LT |
1256 | |
1257 | /* | |
89bddce5 | 1258 | * Check protocol is in range |
1da177e4 LT |
1259 | */ |
1260 | if (family < 0 || family >= NPROTO) | |
1261 | return -EAFNOSUPPORT; | |
1262 | if (type < 0 || type >= SOCK_MAX) | |
1263 | return -EINVAL; | |
1264 | ||
1265 | /* Compatibility. | |
1266 | ||
1267 | This uglymoron is moved from INET layer to here to avoid | |
1268 | deadlock in module load. | |
1269 | */ | |
1270 | if (family == PF_INET && type == SOCK_PACKET) { | |
89bddce5 | 1271 | static int warned; |
1da177e4 LT |
1272 | if (!warned) { |
1273 | warned = 1; | |
89bddce5 SH |
1274 | printk(KERN_INFO "%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1275 | current->comm); | |
1da177e4 LT |
1276 | } |
1277 | family = PF_PACKET; | |
1278 | } | |
1279 | ||
1280 | err = security_socket_create(family, type, protocol, kern); | |
1281 | if (err) | |
1282 | return err; | |
89bddce5 | 1283 | |
55737fda SH |
1284 | /* |
1285 | * Allocate the socket and allow the family to set things up. if | |
1286 | * the protocol is 0, the family is instructed to select an appropriate | |
1287 | * default. | |
1288 | */ | |
1289 | sock = sock_alloc(); | |
1290 | if (!sock) { | |
e87cc472 | 1291 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1292 | return -ENFILE; /* Not exactly a match, but its the |
1293 | closest posix thing */ | |
1294 | } | |
1295 | ||
1296 | sock->type = type; | |
1297 | ||
95a5afca | 1298 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1299 | /* Attempt to load a protocol module if the find failed. |
1300 | * | |
1301 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1302 | * requested real, full-featured networking support upon configuration. |
1303 | * Otherwise module support will break! | |
1304 | */ | |
190683a9 | 1305 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1306 | request_module("net-pf-%d", family); |
1da177e4 LT |
1307 | #endif |
1308 | ||
55737fda SH |
1309 | rcu_read_lock(); |
1310 | pf = rcu_dereference(net_families[family]); | |
1311 | err = -EAFNOSUPPORT; | |
1312 | if (!pf) | |
1313 | goto out_release; | |
1da177e4 LT |
1314 | |
1315 | /* | |
1316 | * We will call the ->create function, that possibly is in a loadable | |
1317 | * module, so we have to bump that loadable module refcnt first. | |
1318 | */ | |
55737fda | 1319 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1320 | goto out_release; |
1321 | ||
55737fda SH |
1322 | /* Now protected by module ref count */ |
1323 | rcu_read_unlock(); | |
1324 | ||
3f378b68 | 1325 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1326 | if (err < 0) |
1da177e4 | 1327 | goto out_module_put; |
a79af59e | 1328 | |
1da177e4 LT |
1329 | /* |
1330 | * Now to bump the refcnt of the [loadable] module that owns this | |
1331 | * socket at sock_release time we decrement its refcnt. | |
1332 | */ | |
55737fda SH |
1333 | if (!try_module_get(sock->ops->owner)) |
1334 | goto out_module_busy; | |
1335 | ||
1da177e4 LT |
1336 | /* |
1337 | * Now that we're done with the ->create function, the [loadable] | |
1338 | * module can have its refcnt decremented | |
1339 | */ | |
55737fda | 1340 | module_put(pf->owner); |
7420ed23 VY |
1341 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1342 | if (err) | |
3b185525 | 1343 | goto out_sock_release; |
55737fda | 1344 | *res = sock; |
1da177e4 | 1345 | |
55737fda SH |
1346 | return 0; |
1347 | ||
1348 | out_module_busy: | |
1349 | err = -EAFNOSUPPORT; | |
1da177e4 | 1350 | out_module_put: |
55737fda SH |
1351 | sock->ops = NULL; |
1352 | module_put(pf->owner); | |
1353 | out_sock_release: | |
1da177e4 | 1354 | sock_release(sock); |
55737fda SH |
1355 | return err; |
1356 | ||
1357 | out_release: | |
1358 | rcu_read_unlock(); | |
1359 | goto out_sock_release; | |
1da177e4 | 1360 | } |
721db93a | 1361 | EXPORT_SYMBOL(__sock_create); |
1da177e4 LT |
1362 | |
1363 | int sock_create(int family, int type, int protocol, struct socket **res) | |
1364 | { | |
1b8d7ae4 | 1365 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1366 | } |
c6d409cf | 1367 | EXPORT_SYMBOL(sock_create); |
1da177e4 LT |
1368 | |
1369 | int sock_create_kern(int family, int type, int protocol, struct socket **res) | |
1370 | { | |
1b8d7ae4 | 1371 | return __sock_create(&init_net, family, type, protocol, res, 1); |
1da177e4 | 1372 | } |
c6d409cf | 1373 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1374 | |
3e0fa65f | 1375 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1da177e4 LT |
1376 | { |
1377 | int retval; | |
1378 | struct socket *sock; | |
a677a039 UD |
1379 | int flags; |
1380 | ||
e38b36f3 UD |
1381 | /* Check the SOCK_* constants for consistency. */ |
1382 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1383 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1384 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1385 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1386 | ||
a677a039 | 1387 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1388 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1389 | return -EINVAL; |
1390 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1391 | |
aaca0bdc UD |
1392 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1393 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1394 | ||
1da177e4 LT |
1395 | retval = sock_create(family, type, protocol, &sock); |
1396 | if (retval < 0) | |
1397 | goto out; | |
1398 | ||
77d27200 | 1399 | retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1400 | if (retval < 0) |
1401 | goto out_release; | |
1402 | ||
1403 | out: | |
1404 | /* It may be already another descriptor 8) Not kernel problem. */ | |
1405 | return retval; | |
1406 | ||
1407 | out_release: | |
1408 | sock_release(sock); | |
1409 | return retval; | |
1410 | } | |
1411 | ||
1412 | /* | |
1413 | * Create a pair of connected sockets. | |
1414 | */ | |
1415 | ||
3e0fa65f HC |
1416 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1417 | int __user *, usockvec) | |
1da177e4 LT |
1418 | { |
1419 | struct socket *sock1, *sock2; | |
1420 | int fd1, fd2, err; | |
db349509 | 1421 | struct file *newfile1, *newfile2; |
a677a039 UD |
1422 | int flags; |
1423 | ||
1424 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1425 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1426 | return -EINVAL; |
1427 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1428 | |
aaca0bdc UD |
1429 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1430 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1431 | ||
1da177e4 LT |
1432 | /* |
1433 | * Obtain the first socket and check if the underlying protocol | |
1434 | * supports the socketpair call. | |
1435 | */ | |
1436 | ||
1437 | err = sock_create(family, type, protocol, &sock1); | |
1438 | if (err < 0) | |
1439 | goto out; | |
1440 | ||
1441 | err = sock_create(family, type, protocol, &sock2); | |
1442 | if (err < 0) | |
1443 | goto out_release_1; | |
1444 | ||
1445 | err = sock1->ops->socketpair(sock1, sock2); | |
89bddce5 | 1446 | if (err < 0) |
1da177e4 LT |
1447 | goto out_release_both; |
1448 | ||
28407630 | 1449 | fd1 = get_unused_fd_flags(flags); |
bf3c23d1 DM |
1450 | if (unlikely(fd1 < 0)) { |
1451 | err = fd1; | |
db349509 | 1452 | goto out_release_both; |
bf3c23d1 | 1453 | } |
28407630 | 1454 | fd2 = get_unused_fd_flags(flags); |
198de4d7 AV |
1455 | if (unlikely(fd2 < 0)) { |
1456 | err = fd2; | |
28407630 AV |
1457 | put_unused_fd(fd1); |
1458 | goto out_release_both; | |
1459 | } | |
1460 | ||
aab174f0 | 1461 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
28407630 AV |
1462 | if (unlikely(IS_ERR(newfile1))) { |
1463 | err = PTR_ERR(newfile1); | |
1464 | put_unused_fd(fd1); | |
1465 | put_unused_fd(fd2); | |
1466 | goto out_release_both; | |
1467 | } | |
1468 | ||
aab174f0 | 1469 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1470 | if (IS_ERR(newfile2)) { |
1471 | err = PTR_ERR(newfile2); | |
198de4d7 AV |
1472 | fput(newfile1); |
1473 | put_unused_fd(fd1); | |
28407630 | 1474 | put_unused_fd(fd2); |
198de4d7 AV |
1475 | sock_release(sock2); |
1476 | goto out; | |
db349509 AV |
1477 | } |
1478 | ||
157cf649 | 1479 | audit_fd_pair(fd1, fd2); |
db349509 AV |
1480 | fd_install(fd1, newfile1); |
1481 | fd_install(fd2, newfile2); | |
1da177e4 LT |
1482 | /* fd1 and fd2 may be already another descriptors. |
1483 | * Not kernel problem. | |
1484 | */ | |
1485 | ||
89bddce5 | 1486 | err = put_user(fd1, &usockvec[0]); |
1da177e4 LT |
1487 | if (!err) |
1488 | err = put_user(fd2, &usockvec[1]); | |
1489 | if (!err) | |
1490 | return 0; | |
1491 | ||
1492 | sys_close(fd2); | |
1493 | sys_close(fd1); | |
1494 | return err; | |
1495 | ||
1da177e4 | 1496 | out_release_both: |
89bddce5 | 1497 | sock_release(sock2); |
1da177e4 | 1498 | out_release_1: |
89bddce5 | 1499 | sock_release(sock1); |
1da177e4 LT |
1500 | out: |
1501 | return err; | |
1502 | } | |
1503 | ||
1da177e4 LT |
1504 | /* |
1505 | * Bind a name to a socket. Nothing much to do here since it's | |
1506 | * the protocol's responsibility to handle the local address. | |
1507 | * | |
1508 | * We move the socket address to kernel space before we call | |
1509 | * the protocol layer (having also checked the address is ok). | |
1510 | */ | |
1511 | ||
20f37034 | 1512 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1da177e4 LT |
1513 | { |
1514 | struct socket *sock; | |
230b1839 | 1515 | struct sockaddr_storage address; |
6cb153ca | 1516 | int err, fput_needed; |
1da177e4 | 1517 | |
89bddce5 | 1518 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1519 | if (sock) { |
43db362d | 1520 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
89bddce5 SH |
1521 | if (err >= 0) { |
1522 | err = security_socket_bind(sock, | |
230b1839 | 1523 | (struct sockaddr *)&address, |
89bddce5 | 1524 | addrlen); |
6cb153ca BL |
1525 | if (!err) |
1526 | err = sock->ops->bind(sock, | |
89bddce5 | 1527 | (struct sockaddr *) |
230b1839 | 1528 | &address, addrlen); |
1da177e4 | 1529 | } |
6cb153ca | 1530 | fput_light(sock->file, fput_needed); |
89bddce5 | 1531 | } |
1da177e4 LT |
1532 | return err; |
1533 | } | |
1534 | ||
1da177e4 LT |
1535 | /* |
1536 | * Perform a listen. Basically, we allow the protocol to do anything | |
1537 | * necessary for a listen, and if that works, we mark the socket as | |
1538 | * ready for listening. | |
1539 | */ | |
1540 | ||
3e0fa65f | 1541 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1da177e4 LT |
1542 | { |
1543 | struct socket *sock; | |
6cb153ca | 1544 | int err, fput_needed; |
b8e1f9b5 | 1545 | int somaxconn; |
89bddce5 SH |
1546 | |
1547 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1548 | if (sock) { | |
8efa6e93 | 1549 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1550 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1551 | backlog = somaxconn; |
1da177e4 LT |
1552 | |
1553 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1554 | if (!err) |
1555 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1556 | |
6cb153ca | 1557 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1558 | } |
1559 | return err; | |
1560 | } | |
1561 | ||
1da177e4 LT |
1562 | /* |
1563 | * For accept, we attempt to create a new socket, set up the link | |
1564 | * with the client, wake up the client, then return the new | |
1565 | * connected fd. We collect the address of the connector in kernel | |
1566 | * space and move it to user at the very end. This is unclean because | |
1567 | * we open the socket then return an error. | |
1568 | * | |
1569 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1570 | * status to recvmsg. We need to add that support in a way thats | |
1571 | * clean when we restucture accept also. | |
1572 | */ | |
1573 | ||
20f37034 HC |
1574 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1575 | int __user *, upeer_addrlen, int, flags) | |
1da177e4 LT |
1576 | { |
1577 | struct socket *sock, *newsock; | |
39d8c1b6 | 1578 | struct file *newfile; |
6cb153ca | 1579 | int err, len, newfd, fput_needed; |
230b1839 | 1580 | struct sockaddr_storage address; |
1da177e4 | 1581 | |
77d27200 | 1582 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1583 | return -EINVAL; |
1584 | ||
1585 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1586 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1587 | ||
6cb153ca | 1588 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1589 | if (!sock) |
1590 | goto out; | |
1591 | ||
1592 | err = -ENFILE; | |
c6d409cf ED |
1593 | newsock = sock_alloc(); |
1594 | if (!newsock) | |
1da177e4 LT |
1595 | goto out_put; |
1596 | ||
1597 | newsock->type = sock->type; | |
1598 | newsock->ops = sock->ops; | |
1599 | ||
1da177e4 LT |
1600 | /* |
1601 | * We don't need try_module_get here, as the listening socket (sock) | |
1602 | * has the protocol module (sock->ops->owner) held. | |
1603 | */ | |
1604 | __module_get(newsock->ops->owner); | |
1605 | ||
28407630 | 1606 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1607 | if (unlikely(newfd < 0)) { |
1608 | err = newfd; | |
9a1875e6 DM |
1609 | sock_release(newsock); |
1610 | goto out_put; | |
39d8c1b6 | 1611 | } |
aab174f0 | 1612 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
28407630 AV |
1613 | if (unlikely(IS_ERR(newfile))) { |
1614 | err = PTR_ERR(newfile); | |
1615 | put_unused_fd(newfd); | |
1616 | sock_release(newsock); | |
1617 | goto out_put; | |
1618 | } | |
39d8c1b6 | 1619 | |
a79af59e FF |
1620 | err = security_socket_accept(sock, newsock); |
1621 | if (err) | |
39d8c1b6 | 1622 | goto out_fd; |
a79af59e | 1623 | |
1da177e4 LT |
1624 | err = sock->ops->accept(sock, newsock, sock->file->f_flags); |
1625 | if (err < 0) | |
39d8c1b6 | 1626 | goto out_fd; |
1da177e4 LT |
1627 | |
1628 | if (upeer_sockaddr) { | |
230b1839 | 1629 | if (newsock->ops->getname(newsock, (struct sockaddr *)&address, |
89bddce5 | 1630 | &len, 2) < 0) { |
1da177e4 | 1631 | err = -ECONNABORTED; |
39d8c1b6 | 1632 | goto out_fd; |
1da177e4 | 1633 | } |
43db362d | 1634 | err = move_addr_to_user(&address, |
230b1839 | 1635 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1636 | if (err < 0) |
39d8c1b6 | 1637 | goto out_fd; |
1da177e4 LT |
1638 | } |
1639 | ||
1640 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1641 | ||
39d8c1b6 DM |
1642 | fd_install(newfd, newfile); |
1643 | err = newfd; | |
1da177e4 | 1644 | |
1da177e4 | 1645 | out_put: |
6cb153ca | 1646 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1647 | out: |
1648 | return err; | |
39d8c1b6 | 1649 | out_fd: |
9606a216 | 1650 | fput(newfile); |
39d8c1b6 | 1651 | put_unused_fd(newfd); |
1da177e4 LT |
1652 | goto out_put; |
1653 | } | |
1654 | ||
20f37034 HC |
1655 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1656 | int __user *, upeer_addrlen) | |
aaca0bdc | 1657 | { |
de11defe | 1658 | return sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1659 | } |
1660 | ||
1da177e4 LT |
1661 | /* |
1662 | * Attempt to connect to a socket with the server address. The address | |
1663 | * is in user space so we verify it is OK and move it to kernel space. | |
1664 | * | |
1665 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1666 | * break bindings | |
1667 | * | |
1668 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1669 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1670 | * include the -EINPROGRESS status for such sockets. | |
1671 | */ | |
1672 | ||
20f37034 HC |
1673 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1674 | int, addrlen) | |
1da177e4 LT |
1675 | { |
1676 | struct socket *sock; | |
230b1839 | 1677 | struct sockaddr_storage address; |
6cb153ca | 1678 | int err, fput_needed; |
1da177e4 | 1679 | |
6cb153ca | 1680 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1681 | if (!sock) |
1682 | goto out; | |
43db362d | 1683 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1684 | if (err < 0) |
1685 | goto out_put; | |
1686 | ||
89bddce5 | 1687 | err = |
230b1839 | 1688 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1689 | if (err) |
1690 | goto out_put; | |
1691 | ||
230b1839 | 1692 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1693 | sock->file->f_flags); |
1694 | out_put: | |
6cb153ca | 1695 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1696 | out: |
1697 | return err; | |
1698 | } | |
1699 | ||
1700 | /* | |
1701 | * Get the local address ('name') of a socket object. Move the obtained | |
1702 | * name to user space. | |
1703 | */ | |
1704 | ||
20f37034 HC |
1705 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1706 | int __user *, usockaddr_len) | |
1da177e4 LT |
1707 | { |
1708 | struct socket *sock; | |
230b1839 | 1709 | struct sockaddr_storage address; |
6cb153ca | 1710 | int len, err, fput_needed; |
89bddce5 | 1711 | |
6cb153ca | 1712 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1713 | if (!sock) |
1714 | goto out; | |
1715 | ||
1716 | err = security_socket_getsockname(sock); | |
1717 | if (err) | |
1718 | goto out_put; | |
1719 | ||
230b1839 | 1720 | err = sock->ops->getname(sock, (struct sockaddr *)&address, &len, 0); |
1da177e4 LT |
1721 | if (err) |
1722 | goto out_put; | |
43db362d | 1723 | err = move_addr_to_user(&address, len, usockaddr, usockaddr_len); |
1da177e4 LT |
1724 | |
1725 | out_put: | |
6cb153ca | 1726 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1727 | out: |
1728 | return err; | |
1729 | } | |
1730 | ||
1731 | /* | |
1732 | * Get the remote address ('name') of a socket object. Move the obtained | |
1733 | * name to user space. | |
1734 | */ | |
1735 | ||
20f37034 HC |
1736 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1737 | int __user *, usockaddr_len) | |
1da177e4 LT |
1738 | { |
1739 | struct socket *sock; | |
230b1839 | 1740 | struct sockaddr_storage address; |
6cb153ca | 1741 | int len, err, fput_needed; |
1da177e4 | 1742 | |
89bddce5 SH |
1743 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1744 | if (sock != NULL) { | |
1da177e4 LT |
1745 | err = security_socket_getpeername(sock); |
1746 | if (err) { | |
6cb153ca | 1747 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1748 | return err; |
1749 | } | |
1750 | ||
89bddce5 | 1751 | err = |
230b1839 | 1752 | sock->ops->getname(sock, (struct sockaddr *)&address, &len, |
89bddce5 | 1753 | 1); |
1da177e4 | 1754 | if (!err) |
43db362d | 1755 | err = move_addr_to_user(&address, len, usockaddr, |
89bddce5 | 1756 | usockaddr_len); |
6cb153ca | 1757 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1758 | } |
1759 | return err; | |
1760 | } | |
1761 | ||
1762 | /* | |
1763 | * Send a datagram to a given address. We move the address into kernel | |
1764 | * space and check the user space data area is readable before invoking | |
1765 | * the protocol. | |
1766 | */ | |
1767 | ||
3e0fa65f | 1768 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1769 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1770 | int, addr_len) |
1da177e4 LT |
1771 | { |
1772 | struct socket *sock; | |
230b1839 | 1773 | struct sockaddr_storage address; |
1da177e4 LT |
1774 | int err; |
1775 | struct msghdr msg; | |
1776 | struct iovec iov; | |
6cb153ca | 1777 | int fput_needed; |
6cb153ca | 1778 | |
253eacc0 LT |
1779 | if (len > INT_MAX) |
1780 | len = INT_MAX; | |
3c2a0909 S |
1781 | if (unlikely(!access_ok(VERIFY_READ, buff, len))) |
1782 | return -EFAULT; | |
de0fa95c PE |
1783 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1784 | if (!sock) | |
4387ff75 | 1785 | goto out; |
6cb153ca | 1786 | |
89bddce5 SH |
1787 | iov.iov_base = buff; |
1788 | iov.iov_len = len; | |
1789 | msg.msg_name = NULL; | |
1790 | msg.msg_iov = &iov; | |
1791 | msg.msg_iovlen = 1; | |
1792 | msg.msg_control = NULL; | |
1793 | msg.msg_controllen = 0; | |
1794 | msg.msg_namelen = 0; | |
6cb153ca | 1795 | if (addr) { |
43db362d | 1796 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1797 | if (err < 0) |
1798 | goto out_put; | |
230b1839 | 1799 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1800 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1801 | } |
1802 | if (sock->file->f_flags & O_NONBLOCK) | |
1803 | flags |= MSG_DONTWAIT; | |
1804 | msg.msg_flags = flags; | |
1805 | err = sock_sendmsg(sock, &msg, len); | |
1806 | ||
89bddce5 | 1807 | out_put: |
de0fa95c | 1808 | fput_light(sock->file, fput_needed); |
4387ff75 | 1809 | out: |
1da177e4 LT |
1810 | return err; |
1811 | } | |
1812 | ||
1813 | /* | |
89bddce5 | 1814 | * Send a datagram down a socket. |
1da177e4 LT |
1815 | */ |
1816 | ||
3e0fa65f | 1817 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1818 | unsigned int, flags) |
1da177e4 LT |
1819 | { |
1820 | return sys_sendto(fd, buff, len, flags, NULL, 0); | |
1821 | } | |
1822 | ||
1823 | /* | |
89bddce5 | 1824 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1825 | * sender. We verify the buffers are writable and if needed move the |
1826 | * sender address from kernel to user space. | |
1827 | */ | |
1828 | ||
3e0fa65f | 1829 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
95c96174 | 1830 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1831 | int __user *, addr_len) |
1da177e4 LT |
1832 | { |
1833 | struct socket *sock; | |
1834 | struct iovec iov; | |
1835 | struct msghdr msg; | |
230b1839 | 1836 | struct sockaddr_storage address; |
89bddce5 | 1837 | int err, err2; |
6cb153ca BL |
1838 | int fput_needed; |
1839 | ||
253eacc0 LT |
1840 | if (size > INT_MAX) |
1841 | size = INT_MAX; | |
3c2a0909 S |
1842 | if (unlikely(!access_ok(VERIFY_WRITE, ubuf, size))) |
1843 | return -EFAULT; | |
de0fa95c | 1844 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 1845 | if (!sock) |
de0fa95c | 1846 | goto out; |
1da177e4 | 1847 | |
89bddce5 SH |
1848 | msg.msg_control = NULL; |
1849 | msg.msg_controllen = 0; | |
1850 | msg.msg_iovlen = 1; | |
1851 | msg.msg_iov = &iov; | |
1852 | iov.iov_len = size; | |
1853 | iov.iov_base = ubuf; | |
2f73d7fd HFS |
1854 | /* Save some cycles and don't copy the address if not needed */ |
1855 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
1856 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
1857 | msg.msg_namelen = 0; | |
1da177e4 LT |
1858 | if (sock->file->f_flags & O_NONBLOCK) |
1859 | flags |= MSG_DONTWAIT; | |
89bddce5 | 1860 | err = sock_recvmsg(sock, &msg, size, flags); |
1da177e4 | 1861 | |
89bddce5 | 1862 | if (err >= 0 && addr != NULL) { |
43db362d | 1863 | err2 = move_addr_to_user(&address, |
230b1839 | 1864 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
1865 | if (err2 < 0) |
1866 | err = err2; | |
1da177e4 | 1867 | } |
de0fa95c PE |
1868 | |
1869 | fput_light(sock->file, fput_needed); | |
4387ff75 | 1870 | out: |
1da177e4 LT |
1871 | return err; |
1872 | } | |
1873 | ||
1874 | /* | |
89bddce5 | 1875 | * Receive a datagram from a socket. |
1da177e4 LT |
1876 | */ |
1877 | ||
89bddce5 | 1878 | asmlinkage long sys_recv(int fd, void __user *ubuf, size_t size, |
95c96174 | 1879 | unsigned int flags) |
1da177e4 LT |
1880 | { |
1881 | return sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); | |
1882 | } | |
1883 | ||
1884 | /* | |
1885 | * Set a socket option. Because we don't know the option lengths we have | |
1886 | * to pass the user mode parameter for the protocols to sort out. | |
1887 | */ | |
1888 | ||
20f37034 HC |
1889 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
1890 | char __user *, optval, int, optlen) | |
1da177e4 | 1891 | { |
6cb153ca | 1892 | int err, fput_needed; |
1da177e4 LT |
1893 | struct socket *sock; |
1894 | ||
1895 | if (optlen < 0) | |
1896 | return -EINVAL; | |
89bddce5 SH |
1897 | |
1898 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1899 | if (sock != NULL) { | |
1900 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
1901 | if (err) |
1902 | goto out_put; | |
1da177e4 LT |
1903 | |
1904 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1905 | err = |
1906 | sock_setsockopt(sock, level, optname, optval, | |
1907 | optlen); | |
1da177e4 | 1908 | else |
89bddce5 SH |
1909 | err = |
1910 | sock->ops->setsockopt(sock, level, optname, optval, | |
1911 | optlen); | |
6cb153ca BL |
1912 | out_put: |
1913 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1914 | } |
1915 | return err; | |
1916 | } | |
1917 | ||
1918 | /* | |
1919 | * Get a socket option. Because we don't know the option lengths we have | |
1920 | * to pass a user mode parameter for the protocols to sort out. | |
1921 | */ | |
1922 | ||
20f37034 HC |
1923 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
1924 | char __user *, optval, int __user *, optlen) | |
1da177e4 | 1925 | { |
6cb153ca | 1926 | int err, fput_needed; |
1da177e4 LT |
1927 | struct socket *sock; |
1928 | ||
89bddce5 SH |
1929 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1930 | if (sock != NULL) { | |
6cb153ca BL |
1931 | err = security_socket_getsockopt(sock, level, optname); |
1932 | if (err) | |
1933 | goto out_put; | |
1da177e4 LT |
1934 | |
1935 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1936 | err = |
1937 | sock_getsockopt(sock, level, optname, optval, | |
1938 | optlen); | |
1da177e4 | 1939 | else |
89bddce5 SH |
1940 | err = |
1941 | sock->ops->getsockopt(sock, level, optname, optval, | |
1942 | optlen); | |
6cb153ca BL |
1943 | out_put: |
1944 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1945 | } |
1946 | return err; | |
1947 | } | |
1948 | ||
1da177e4 LT |
1949 | /* |
1950 | * Shutdown a socket. | |
1951 | */ | |
1952 | ||
754fe8d2 | 1953 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
1da177e4 | 1954 | { |
6cb153ca | 1955 | int err, fput_needed; |
1da177e4 LT |
1956 | struct socket *sock; |
1957 | ||
89bddce5 SH |
1958 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1959 | if (sock != NULL) { | |
1da177e4 | 1960 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
1961 | if (!err) |
1962 | err = sock->ops->shutdown(sock, how); | |
1963 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1964 | } |
1965 | return err; | |
1966 | } | |
1967 | ||
89bddce5 | 1968 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
1969 | * fields which are the same type (int / unsigned) on our platforms. |
1970 | */ | |
1971 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
1972 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
1973 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
1974 | ||
c71d8ebe TH |
1975 | struct used_address { |
1976 | struct sockaddr_storage name; | |
1977 | unsigned int name_len; | |
1978 | }; | |
1979 | ||
2e8d97ab DC |
1980 | static int copy_msghdr_from_user(struct msghdr *kmsg, |
1981 | struct msghdr __user *umsg) | |
1982 | { | |
1983 | if (copy_from_user(kmsg, umsg, sizeof(struct msghdr))) | |
1984 | return -EFAULT; | |
29a322f4 ML |
1985 | |
1986 | if (kmsg->msg_namelen < 0) | |
1987 | return -EINVAL; | |
1988 | ||
2e8d97ab | 1989 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
ed47bba2 | 1990 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
2e8d97ab DC |
1991 | return 0; |
1992 | } | |
1993 | ||
a7526eb5 | 1994 | static int ___sys_sendmsg(struct socket *sock, struct msghdr __user *msg, |
95c96174 | 1995 | struct msghdr *msg_sys, unsigned int flags, |
c71d8ebe | 1996 | struct used_address *used_address) |
1da177e4 | 1997 | { |
89bddce5 SH |
1998 | struct compat_msghdr __user *msg_compat = |
1999 | (struct compat_msghdr __user *)msg; | |
230b1839 | 2000 | struct sockaddr_storage address; |
1da177e4 | 2001 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 2002 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
89bddce5 SH |
2003 | __attribute__ ((aligned(sizeof(__kernel_size_t)))); |
2004 | /* 20 is size of ipv6_pktinfo */ | |
1da177e4 | 2005 | unsigned char *ctl_buf = ctl; |
a74e9106 | 2006 | int err, ctl_len, total_len; |
89bddce5 | 2007 | |
1da177e4 | 2008 | err = -EFAULT; |
2481f1e9 JZ |
2009 | if (MSG_CMSG_COMPAT & flags) |
2010 | err = get_compat_msghdr(msg_sys, msg_compat); | |
2011 | else | |
2e8d97ab | 2012 | err = copy_msghdr_from_user(msg_sys, msg); |
2481f1e9 JZ |
2013 | if (err) |
2014 | return err; | |
1da177e4 | 2015 | |
228e548e | 2016 | if (msg_sys->msg_iovlen > UIO_FASTIOV) { |
a74e9106 ED |
2017 | err = -EMSGSIZE; |
2018 | if (msg_sys->msg_iovlen > UIO_MAXIOV) | |
2019 | goto out; | |
2020 | err = -ENOMEM; | |
2021 | iov = kmalloc(msg_sys->msg_iovlen * sizeof(struct iovec), | |
2022 | GFP_KERNEL); | |
1da177e4 | 2023 | if (!iov) |
228e548e | 2024 | goto out; |
1da177e4 LT |
2025 | } |
2026 | ||
2027 | /* This will also move the address data into kernel space */ | |
2028 | if (MSG_CMSG_COMPAT & flags) { | |
43db362d | 2029 | err = verify_compat_iovec(msg_sys, iov, &address, VERIFY_READ); |
1da177e4 | 2030 | } else |
43db362d | 2031 | err = verify_iovec(msg_sys, iov, &address, VERIFY_READ); |
89bddce5 | 2032 | if (err < 0) |
1da177e4 LT |
2033 | goto out_freeiov; |
2034 | total_len = err; | |
2035 | ||
2036 | err = -ENOBUFS; | |
2037 | ||
228e548e | 2038 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 2039 | goto out_freeiov; |
228e548e | 2040 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2041 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2042 | err = |
228e548e | 2043 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2044 | sizeof(ctl)); |
1da177e4 LT |
2045 | if (err) |
2046 | goto out_freeiov; | |
228e548e AB |
2047 | ctl_buf = msg_sys->msg_control; |
2048 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2049 | } else if (ctl_len) { |
89bddce5 | 2050 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2051 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2052 | if (ctl_buf == NULL) |
1da177e4 LT |
2053 | goto out_freeiov; |
2054 | } | |
2055 | err = -EFAULT; | |
2056 | /* | |
228e548e | 2057 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2058 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2059 | * checking falls down on this. | |
2060 | */ | |
fb8621bb | 2061 | if (copy_from_user(ctl_buf, |
228e548e | 2062 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2063 | ctl_len)) |
1da177e4 | 2064 | goto out_freectl; |
228e548e | 2065 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2066 | } |
228e548e | 2067 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2068 | |
2069 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2070 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2071 | /* |
2072 | * If this is sendmmsg() and current destination address is same as | |
2073 | * previously succeeded address, omit asking LSM's decision. | |
2074 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2075 | * destination address never matches. | |
2076 | */ | |
bc909d9d MD |
2077 | if (used_address && msg_sys->msg_name && |
2078 | used_address->name_len == msg_sys->msg_namelen && | |
2079 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe TH |
2080 | used_address->name_len)) { |
2081 | err = sock_sendmsg_nosec(sock, msg_sys, total_len); | |
2082 | goto out_freectl; | |
2083 | } | |
2084 | err = sock_sendmsg(sock, msg_sys, total_len); | |
2085 | /* | |
2086 | * If this is sendmmsg() and sending to current destination address was | |
2087 | * successful, remember it. | |
2088 | */ | |
2089 | if (used_address && err >= 0) { | |
2090 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2091 | if (msg_sys->msg_name) |
2092 | memcpy(&used_address->name, msg_sys->msg_name, | |
2093 | used_address->name_len); | |
c71d8ebe | 2094 | } |
1da177e4 LT |
2095 | |
2096 | out_freectl: | |
89bddce5 | 2097 | if (ctl_buf != ctl) |
1da177e4 LT |
2098 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
2099 | out_freeiov: | |
2100 | if (iov != iovstack) | |
a74e9106 | 2101 | kfree(iov); |
228e548e AB |
2102 | out: |
2103 | return err; | |
2104 | } | |
2105 | ||
2106 | /* | |
2107 | * BSD sendmsg interface | |
2108 | */ | |
2109 | ||
a7526eb5 | 2110 | long __sys_sendmsg(int fd, struct msghdr __user *msg, unsigned flags) |
228e548e AB |
2111 | { |
2112 | int fput_needed, err; | |
2113 | struct msghdr msg_sys; | |
1be374a0 AL |
2114 | struct socket *sock; |
2115 | ||
1be374a0 | 2116 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2117 | if (!sock) |
2118 | goto out; | |
2119 | ||
a7526eb5 | 2120 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL); |
228e548e | 2121 | |
6cb153ca | 2122 | fput_light(sock->file, fput_needed); |
89bddce5 | 2123 | out: |
1da177e4 LT |
2124 | return err; |
2125 | } | |
2126 | ||
a7526eb5 AL |
2127 | SYSCALL_DEFINE3(sendmsg, int, fd, struct msghdr __user *, msg, unsigned int, flags) |
2128 | { | |
2129 | if (flags & MSG_CMSG_COMPAT) | |
2130 | return -EINVAL; | |
2131 | return __sys_sendmsg(fd, msg, flags); | |
2132 | } | |
2133 | ||
228e548e AB |
2134 | /* |
2135 | * Linux sendmmsg interface | |
2136 | */ | |
2137 | ||
2138 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2139 | unsigned int flags) | |
2140 | { | |
2141 | int fput_needed, err, datagrams; | |
2142 | struct socket *sock; | |
2143 | struct mmsghdr __user *entry; | |
2144 | struct compat_mmsghdr __user *compat_entry; | |
2145 | struct msghdr msg_sys; | |
c71d8ebe | 2146 | struct used_address used_address; |
228e548e | 2147 | |
98382f41 AB |
2148 | if (vlen > UIO_MAXIOV) |
2149 | vlen = UIO_MAXIOV; | |
228e548e AB |
2150 | |
2151 | datagrams = 0; | |
2152 | ||
2153 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2154 | if (!sock) | |
2155 | return err; | |
2156 | ||
c71d8ebe | 2157 | used_address.name_len = UINT_MAX; |
228e548e AB |
2158 | entry = mmsg; |
2159 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2160 | err = 0; |
228e548e AB |
2161 | |
2162 | while (datagrams < vlen) { | |
228e548e | 2163 | if (MSG_CMSG_COMPAT & flags) { |
a7526eb5 AL |
2164 | err = ___sys_sendmsg(sock, (struct msghdr __user *)compat_entry, |
2165 | &msg_sys, flags, &used_address); | |
228e548e AB |
2166 | if (err < 0) |
2167 | break; | |
2168 | err = __put_user(err, &compat_entry->msg_len); | |
2169 | ++compat_entry; | |
2170 | } else { | |
a7526eb5 AL |
2171 | err = ___sys_sendmsg(sock, |
2172 | (struct msghdr __user *)entry, | |
2173 | &msg_sys, flags, &used_address); | |
228e548e AB |
2174 | if (err < 0) |
2175 | break; | |
2176 | err = put_user(err, &entry->msg_len); | |
2177 | ++entry; | |
2178 | } | |
2179 | ||
2180 | if (err) | |
2181 | break; | |
2182 | ++datagrams; | |
2183 | } | |
2184 | ||
228e548e AB |
2185 | fput_light(sock->file, fput_needed); |
2186 | ||
728ffb86 AB |
2187 | /* We only return an error if no datagrams were able to be sent */ |
2188 | if (datagrams != 0) | |
228e548e AB |
2189 | return datagrams; |
2190 | ||
228e548e AB |
2191 | return err; |
2192 | } | |
2193 | ||
2194 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2195 | unsigned int, vlen, unsigned int, flags) | |
2196 | { | |
1be374a0 AL |
2197 | if (flags & MSG_CMSG_COMPAT) |
2198 | return -EINVAL; | |
228e548e AB |
2199 | return __sys_sendmmsg(fd, mmsg, vlen, flags); |
2200 | } | |
2201 | ||
a7526eb5 | 2202 | static int ___sys_recvmsg(struct socket *sock, struct msghdr __user *msg, |
95c96174 | 2203 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2204 | { |
89bddce5 SH |
2205 | struct compat_msghdr __user *msg_compat = |
2206 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2207 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2208 | struct iovec *iov = iovstack; |
1da177e4 | 2209 | unsigned long cmsg_ptr; |
a74e9106 | 2210 | int err, total_len, len; |
1da177e4 LT |
2211 | |
2212 | /* kernel mode address */ | |
230b1839 | 2213 | struct sockaddr_storage addr; |
1da177e4 LT |
2214 | |
2215 | /* user mode address pointers */ | |
2216 | struct sockaddr __user *uaddr; | |
2217 | int __user *uaddr_len; | |
89bddce5 | 2218 | |
2481f1e9 JZ |
2219 | if (MSG_CMSG_COMPAT & flags) |
2220 | err = get_compat_msghdr(msg_sys, msg_compat); | |
2221 | else | |
2e8d97ab | 2222 | err = copy_msghdr_from_user(msg_sys, msg); |
2481f1e9 JZ |
2223 | if (err) |
2224 | return err; | |
1da177e4 | 2225 | |
a2e27255 | 2226 | if (msg_sys->msg_iovlen > UIO_FASTIOV) { |
a74e9106 ED |
2227 | err = -EMSGSIZE; |
2228 | if (msg_sys->msg_iovlen > UIO_MAXIOV) | |
2229 | goto out; | |
2230 | err = -ENOMEM; | |
2231 | iov = kmalloc(msg_sys->msg_iovlen * sizeof(struct iovec), | |
2232 | GFP_KERNEL); | |
1da177e4 | 2233 | if (!iov) |
a2e27255 | 2234 | goto out; |
1da177e4 LT |
2235 | } |
2236 | ||
2f73d7fd HFS |
2237 | /* Save the user-mode address (verify_iovec will change the |
2238 | * kernel msghdr to use the kernel address space) | |
1da177e4 | 2239 | */ |
a2e27255 | 2240 | uaddr = (__force void __user *)msg_sys->msg_name; |
1da177e4 | 2241 | uaddr_len = COMPAT_NAMELEN(msg); |
2f73d7fd | 2242 | if (MSG_CMSG_COMPAT & flags) |
43db362d | 2243 | err = verify_compat_iovec(msg_sys, iov, &addr, VERIFY_WRITE); |
2f73d7fd | 2244 | else |
43db362d | 2245 | err = verify_iovec(msg_sys, iov, &addr, VERIFY_WRITE); |
1da177e4 LT |
2246 | if (err < 0) |
2247 | goto out_freeiov; | |
89bddce5 | 2248 | total_len = err; |
1da177e4 | 2249 | |
a2e27255 ACM |
2250 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2251 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2252 | |
2f73d7fd HFS |
2253 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2254 | msg_sys->msg_namelen = 0; | |
2255 | ||
1da177e4 LT |
2256 | if (sock->file->f_flags & O_NONBLOCK) |
2257 | flags |= MSG_DONTWAIT; | |
a2e27255 ACM |
2258 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, |
2259 | total_len, flags); | |
1da177e4 LT |
2260 | if (err < 0) |
2261 | goto out_freeiov; | |
2262 | len = err; | |
2263 | ||
2264 | if (uaddr != NULL) { | |
43db362d | 2265 | err = move_addr_to_user(&addr, |
a2e27255 | 2266 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2267 | uaddr_len); |
1da177e4 LT |
2268 | if (err < 0) |
2269 | goto out_freeiov; | |
2270 | } | |
a2e27255 | 2271 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2272 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2273 | if (err) |
2274 | goto out_freeiov; | |
2275 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2276 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2277 | &msg_compat->msg_controllen); |
2278 | else | |
a2e27255 | 2279 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2280 | &msg->msg_controllen); |
2281 | if (err) | |
2282 | goto out_freeiov; | |
2283 | err = len; | |
2284 | ||
2285 | out_freeiov: | |
2286 | if (iov != iovstack) | |
a74e9106 | 2287 | kfree(iov); |
a2e27255 ACM |
2288 | out: |
2289 | return err; | |
2290 | } | |
2291 | ||
2292 | /* | |
2293 | * BSD recvmsg interface | |
2294 | */ | |
2295 | ||
a7526eb5 | 2296 | long __sys_recvmsg(int fd, struct msghdr __user *msg, unsigned flags) |
a2e27255 ACM |
2297 | { |
2298 | int fput_needed, err; | |
2299 | struct msghdr msg_sys; | |
1be374a0 AL |
2300 | struct socket *sock; |
2301 | ||
1be374a0 | 2302 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2303 | if (!sock) |
2304 | goto out; | |
2305 | ||
a7526eb5 | 2306 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2307 | |
6cb153ca | 2308 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2309 | out: |
2310 | return err; | |
2311 | } | |
2312 | ||
a7526eb5 AL |
2313 | SYSCALL_DEFINE3(recvmsg, int, fd, struct msghdr __user *, msg, |
2314 | unsigned int, flags) | |
2315 | { | |
2316 | if (flags & MSG_CMSG_COMPAT) | |
2317 | return -EINVAL; | |
2318 | return __sys_recvmsg(fd, msg, flags); | |
2319 | } | |
2320 | ||
a2e27255 ACM |
2321 | /* |
2322 | * Linux recvmmsg interface | |
2323 | */ | |
2324 | ||
2325 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2326 | unsigned int flags, struct timespec *timeout) | |
2327 | { | |
2328 | int fput_needed, err, datagrams; | |
2329 | struct socket *sock; | |
2330 | struct mmsghdr __user *entry; | |
d7256d0e | 2331 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 ACM |
2332 | struct msghdr msg_sys; |
2333 | struct timespec end_time; | |
2334 | ||
2335 | if (timeout && | |
2336 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2337 | timeout->tv_nsec)) | |
2338 | return -EINVAL; | |
2339 | ||
2340 | datagrams = 0; | |
2341 | ||
2342 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2343 | if (!sock) | |
2344 | return err; | |
2345 | ||
2346 | err = sock_error(sock->sk); | |
90a1c226 MJ |
2347 | if (err) { |
2348 | datagrams = err; | |
a2e27255 | 2349 | goto out_put; |
90a1c226 | 2350 | } |
a2e27255 ACM |
2351 | |
2352 | entry = mmsg; | |
d7256d0e | 2353 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2354 | |
2355 | while (datagrams < vlen) { | |
2356 | /* | |
2357 | * No need to ask LSM for more than the first datagram. | |
2358 | */ | |
d7256d0e | 2359 | if (MSG_CMSG_COMPAT & flags) { |
a7526eb5 AL |
2360 | err = ___sys_recvmsg(sock, (struct msghdr __user *)compat_entry, |
2361 | &msg_sys, flags & ~MSG_WAITFORONE, | |
2362 | datagrams); | |
d7256d0e JMG |
2363 | if (err < 0) |
2364 | break; | |
2365 | err = __put_user(err, &compat_entry->msg_len); | |
2366 | ++compat_entry; | |
2367 | } else { | |
a7526eb5 AL |
2368 | err = ___sys_recvmsg(sock, |
2369 | (struct msghdr __user *)entry, | |
2370 | &msg_sys, flags & ~MSG_WAITFORONE, | |
2371 | datagrams); | |
d7256d0e JMG |
2372 | if (err < 0) |
2373 | break; | |
2374 | err = put_user(err, &entry->msg_len); | |
2375 | ++entry; | |
2376 | } | |
2377 | ||
a2e27255 ACM |
2378 | if (err) |
2379 | break; | |
a2e27255 ACM |
2380 | ++datagrams; |
2381 | ||
71c5c159 BB |
2382 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2383 | if (flags & MSG_WAITFORONE) | |
2384 | flags |= MSG_DONTWAIT; | |
2385 | ||
a2e27255 ACM |
2386 | if (timeout) { |
2387 | ktime_get_ts(timeout); | |
2388 | *timeout = timespec_sub(end_time, *timeout); | |
2389 | if (timeout->tv_sec < 0) { | |
2390 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2391 | break; | |
2392 | } | |
2393 | ||
2394 | /* Timeout, return less than vlen datagrams */ | |
2395 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2396 | break; | |
2397 | } | |
2398 | ||
2399 | /* Out of band data, return right away */ | |
2400 | if (msg_sys.msg_flags & MSG_OOB) | |
2401 | break; | |
2402 | } | |
2403 | ||
a2e27255 | 2404 | if (err == 0) |
3c2a0909 S |
2405 | goto out_put; |
2406 | ||
2407 | if (datagrams == 0) { | |
2408 | datagrams = err; | |
2409 | goto out_put; | |
2410 | } | |
a2e27255 | 2411 | |
3c2a0909 S |
2412 | /* |
2413 | * We may return less entries than requested (vlen) if the | |
2414 | * sock is non block and there aren't enough datagrams... | |
2415 | */ | |
2416 | if (err != -EAGAIN) { | |
a2e27255 | 2417 | /* |
3c2a0909 S |
2418 | * ... or if recvmsg returns an error after we |
2419 | * received some datagrams, where we record the | |
2420 | * error to return on the next call or if the | |
2421 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2422 | */ |
3c2a0909 | 2423 | sock->sk->sk_err = -err; |
a2e27255 | 2424 | } |
3c2a0909 S |
2425 | out_put: |
2426 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2427 | |
3c2a0909 | 2428 | return datagrams; |
a2e27255 ACM |
2429 | } |
2430 | ||
2431 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2432 | unsigned int, vlen, unsigned int, flags, | |
2433 | struct timespec __user *, timeout) | |
2434 | { | |
2435 | int datagrams; | |
2436 | struct timespec timeout_sys; | |
2437 | ||
1be374a0 AL |
2438 | if (flags & MSG_CMSG_COMPAT) |
2439 | return -EINVAL; | |
2440 | ||
a2e27255 ACM |
2441 | if (!timeout) |
2442 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2443 | ||
2444 | if (copy_from_user(&timeout_sys, timeout, sizeof(timeout_sys))) | |
2445 | return -EFAULT; | |
2446 | ||
2447 | datagrams = __sys_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
2448 | ||
2449 | if (datagrams > 0 && | |
2450 | copy_to_user(timeout, &timeout_sys, sizeof(timeout_sys))) | |
2451 | datagrams = -EFAULT; | |
2452 | ||
2453 | return datagrams; | |
2454 | } | |
2455 | ||
2456 | #ifdef __ARCH_WANT_SYS_SOCKETCALL | |
1da177e4 LT |
2457 | /* Argument list sizes for sys_socketcall */ |
2458 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2459 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2460 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2461 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2462 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2463 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2464 | }; |
2465 | ||
1da177e4 LT |
2466 | #undef AL |
2467 | ||
2468 | /* | |
89bddce5 | 2469 | * System call vectors. |
1da177e4 LT |
2470 | * |
2471 | * Argument checking cleaned up. Saved 20% in size. | |
2472 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2473 | * it is set by the callees. |
1da177e4 LT |
2474 | */ |
2475 | ||
3e0fa65f | 2476 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2477 | { |
2950fa9d | 2478 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2479 | unsigned long a0, a1; |
1da177e4 | 2480 | int err; |
47379052 | 2481 | unsigned int len; |
1da177e4 | 2482 | |
228e548e | 2483 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 LT |
2484 | return -EINVAL; |
2485 | ||
47379052 AV |
2486 | len = nargs[call]; |
2487 | if (len > sizeof(a)) | |
2488 | return -EINVAL; | |
2489 | ||
1da177e4 | 2490 | /* copy_from_user should be SMP safe. */ |
47379052 | 2491 | if (copy_from_user(a, args, len)) |
1da177e4 | 2492 | return -EFAULT; |
3ec3b2fb | 2493 | |
2950fa9d CG |
2494 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2495 | if (err) | |
2496 | return err; | |
3ec3b2fb | 2497 | |
89bddce5 SH |
2498 | a0 = a[0]; |
2499 | a1 = a[1]; | |
2500 | ||
2501 | switch (call) { | |
2502 | case SYS_SOCKET: | |
2503 | err = sys_socket(a0, a1, a[2]); | |
2504 | break; | |
2505 | case SYS_BIND: | |
2506 | err = sys_bind(a0, (struct sockaddr __user *)a1, a[2]); | |
2507 | break; | |
2508 | case SYS_CONNECT: | |
2509 | err = sys_connect(a0, (struct sockaddr __user *)a1, a[2]); | |
2510 | break; | |
2511 | case SYS_LISTEN: | |
2512 | err = sys_listen(a0, a1); | |
2513 | break; | |
2514 | case SYS_ACCEPT: | |
de11defe UD |
2515 | err = sys_accept4(a0, (struct sockaddr __user *)a1, |
2516 | (int __user *)a[2], 0); | |
89bddce5 SH |
2517 | break; |
2518 | case SYS_GETSOCKNAME: | |
2519 | err = | |
2520 | sys_getsockname(a0, (struct sockaddr __user *)a1, | |
2521 | (int __user *)a[2]); | |
2522 | break; | |
2523 | case SYS_GETPEERNAME: | |
2524 | err = | |
2525 | sys_getpeername(a0, (struct sockaddr __user *)a1, | |
2526 | (int __user *)a[2]); | |
2527 | break; | |
2528 | case SYS_SOCKETPAIR: | |
2529 | err = sys_socketpair(a0, a1, a[2], (int __user *)a[3]); | |
2530 | break; | |
2531 | case SYS_SEND: | |
2532 | err = sys_send(a0, (void __user *)a1, a[2], a[3]); | |
2533 | break; | |
2534 | case SYS_SENDTO: | |
2535 | err = sys_sendto(a0, (void __user *)a1, a[2], a[3], | |
2536 | (struct sockaddr __user *)a[4], a[5]); | |
2537 | break; | |
2538 | case SYS_RECV: | |
2539 | err = sys_recv(a0, (void __user *)a1, a[2], a[3]); | |
2540 | break; | |
2541 | case SYS_RECVFROM: | |
2542 | err = sys_recvfrom(a0, (void __user *)a1, a[2], a[3], | |
2543 | (struct sockaddr __user *)a[4], | |
2544 | (int __user *)a[5]); | |
2545 | break; | |
2546 | case SYS_SHUTDOWN: | |
2547 | err = sys_shutdown(a0, a1); | |
2548 | break; | |
2549 | case SYS_SETSOCKOPT: | |
2550 | err = sys_setsockopt(a0, a1, a[2], (char __user *)a[3], a[4]); | |
2551 | break; | |
2552 | case SYS_GETSOCKOPT: | |
2553 | err = | |
2554 | sys_getsockopt(a0, a1, a[2], (char __user *)a[3], | |
2555 | (int __user *)a[4]); | |
2556 | break; | |
2557 | case SYS_SENDMSG: | |
2558 | err = sys_sendmsg(a0, (struct msghdr __user *)a1, a[2]); | |
2559 | break; | |
228e548e AB |
2560 | case SYS_SENDMMSG: |
2561 | err = sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3]); | |
2562 | break; | |
89bddce5 SH |
2563 | case SYS_RECVMSG: |
2564 | err = sys_recvmsg(a0, (struct msghdr __user *)a1, a[2]); | |
2565 | break; | |
a2e27255 ACM |
2566 | case SYS_RECVMMSG: |
2567 | err = sys_recvmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3], | |
2568 | (struct timespec __user *)a[4]); | |
2569 | break; | |
de11defe UD |
2570 | case SYS_ACCEPT4: |
2571 | err = sys_accept4(a0, (struct sockaddr __user *)a1, | |
2572 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2573 | break; |
89bddce5 SH |
2574 | default: |
2575 | err = -EINVAL; | |
2576 | break; | |
1da177e4 LT |
2577 | } |
2578 | return err; | |
2579 | } | |
2580 | ||
89bddce5 | 2581 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2582 | |
55737fda SH |
2583 | /** |
2584 | * sock_register - add a socket protocol handler | |
2585 | * @ops: description of protocol | |
2586 | * | |
1da177e4 LT |
2587 | * This function is called by a protocol handler that wants to |
2588 | * advertise its address family, and have it linked into the | |
55737fda SH |
2589 | * socket interface. The value ops->family coresponds to the |
2590 | * socket system call protocol family. | |
1da177e4 | 2591 | */ |
f0fd27d4 | 2592 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2593 | { |
2594 | int err; | |
2595 | ||
2596 | if (ops->family >= NPROTO) { | |
89bddce5 SH |
2597 | printk(KERN_CRIT "protocol %d >= NPROTO(%d)\n", ops->family, |
2598 | NPROTO); | |
1da177e4 LT |
2599 | return -ENOBUFS; |
2600 | } | |
55737fda SH |
2601 | |
2602 | spin_lock(&net_family_lock); | |
190683a9 ED |
2603 | if (rcu_dereference_protected(net_families[ops->family], |
2604 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2605 | err = -EEXIST; |
2606 | else { | |
cf778b00 | 2607 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2608 | err = 0; |
2609 | } | |
55737fda SH |
2610 | spin_unlock(&net_family_lock); |
2611 | ||
89bddce5 | 2612 | printk(KERN_INFO "NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2613 | return err; |
2614 | } | |
c6d409cf | 2615 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2616 | |
55737fda SH |
2617 | /** |
2618 | * sock_unregister - remove a protocol handler | |
2619 | * @family: protocol family to remove | |
2620 | * | |
1da177e4 LT |
2621 | * This function is called by a protocol handler that wants to |
2622 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2623 | * new socket creation. |
2624 | * | |
2625 | * If protocol handler is a module, then it can use module reference | |
2626 | * counts to protect against new references. If protocol handler is not | |
2627 | * a module then it needs to provide its own protection in | |
2628 | * the ops->create routine. | |
1da177e4 | 2629 | */ |
f0fd27d4 | 2630 | void sock_unregister(int family) |
1da177e4 | 2631 | { |
f0fd27d4 | 2632 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2633 | |
55737fda | 2634 | spin_lock(&net_family_lock); |
a9b3cd7f | 2635 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2636 | spin_unlock(&net_family_lock); |
2637 | ||
2638 | synchronize_rcu(); | |
2639 | ||
89bddce5 | 2640 | printk(KERN_INFO "NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2641 | } |
c6d409cf | 2642 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2643 | |
77d76ea3 | 2644 | static int __init sock_init(void) |
1da177e4 | 2645 | { |
b3e19d92 | 2646 | int err; |
2ca794e5 EB |
2647 | /* |
2648 | * Initialize the network sysctl infrastructure. | |
2649 | */ | |
2650 | err = net_sysctl_init(); | |
2651 | if (err) | |
2652 | goto out; | |
b3e19d92 | 2653 | |
1da177e4 | 2654 | /* |
89bddce5 | 2655 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2656 | */ |
2657 | skb_init(); | |
1da177e4 LT |
2658 | |
2659 | /* | |
89bddce5 | 2660 | * Initialize the protocols module. |
1da177e4 LT |
2661 | */ |
2662 | ||
2663 | init_inodecache(); | |
b3e19d92 NP |
2664 | |
2665 | err = register_filesystem(&sock_fs_type); | |
2666 | if (err) | |
2667 | goto out_fs; | |
1da177e4 | 2668 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2669 | if (IS_ERR(sock_mnt)) { |
2670 | err = PTR_ERR(sock_mnt); | |
2671 | goto out_mount; | |
2672 | } | |
77d76ea3 AK |
2673 | |
2674 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2675 | */ |
2676 | ||
2677 | #ifdef CONFIG_NETFILTER | |
2678 | netfilter_init(); | |
2679 | #endif | |
cbeb321a | 2680 | |
c1f19b51 RC |
2681 | #ifdef CONFIG_NETWORK_PHY_TIMESTAMPING |
2682 | skb_timestamping_init(); | |
2683 | #endif | |
2684 | ||
b3e19d92 NP |
2685 | out: |
2686 | return err; | |
2687 | ||
2688 | out_mount: | |
2689 | unregister_filesystem(&sock_fs_type); | |
2690 | out_fs: | |
2691 | goto out; | |
1da177e4 LT |
2692 | } |
2693 | ||
77d76ea3 AK |
2694 | core_initcall(sock_init); /* early initcall */ |
2695 | ||
1da177e4 LT |
2696 | #ifdef CONFIG_PROC_FS |
2697 | void socket_seq_show(struct seq_file *seq) | |
2698 | { | |
2699 | int cpu; | |
2700 | int counter = 0; | |
2701 | ||
6f912042 | 2702 | for_each_possible_cpu(cpu) |
89bddce5 | 2703 | counter += per_cpu(sockets_in_use, cpu); |
1da177e4 LT |
2704 | |
2705 | /* It can be negative, by the way. 8) */ | |
2706 | if (counter < 0) | |
2707 | counter = 0; | |
2708 | ||
2709 | seq_printf(seq, "sockets: used %d\n", counter); | |
2710 | } | |
89bddce5 | 2711 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 2712 | |
89bbfc95 | 2713 | #ifdef CONFIG_COMPAT |
6b96018b | 2714 | static int do_siocgstamp(struct net *net, struct socket *sock, |
644595f8 | 2715 | unsigned int cmd, void __user *up) |
7a229387 | 2716 | { |
7a229387 AB |
2717 | mm_segment_t old_fs = get_fs(); |
2718 | struct timeval ktv; | |
2719 | int err; | |
2720 | ||
2721 | set_fs(KERNEL_DS); | |
6b96018b | 2722 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&ktv); |
7a229387 | 2723 | set_fs(old_fs); |
644595f8 | 2724 | if (!err) |
ed6fe9d6 | 2725 | err = compat_put_timeval(&ktv, up); |
644595f8 | 2726 | |
7a229387 AB |
2727 | return err; |
2728 | } | |
2729 | ||
6b96018b | 2730 | static int do_siocgstampns(struct net *net, struct socket *sock, |
644595f8 | 2731 | unsigned int cmd, void __user *up) |
7a229387 | 2732 | { |
7a229387 AB |
2733 | mm_segment_t old_fs = get_fs(); |
2734 | struct timespec kts; | |
2735 | int err; | |
2736 | ||
2737 | set_fs(KERNEL_DS); | |
6b96018b | 2738 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&kts); |
7a229387 | 2739 | set_fs(old_fs); |
644595f8 | 2740 | if (!err) |
ed6fe9d6 | 2741 | err = compat_put_timespec(&kts, up); |
644595f8 | 2742 | |
7a229387 AB |
2743 | return err; |
2744 | } | |
2745 | ||
6b96018b | 2746 | static int dev_ifname32(struct net *net, struct compat_ifreq __user *uifr32) |
7a229387 AB |
2747 | { |
2748 | struct ifreq __user *uifr; | |
2749 | int err; | |
2750 | ||
2751 | uifr = compat_alloc_user_space(sizeof(struct ifreq)); | |
6b96018b | 2752 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2753 | return -EFAULT; |
2754 | ||
6b96018b | 2755 | err = dev_ioctl(net, SIOCGIFNAME, uifr); |
7a229387 AB |
2756 | if (err) |
2757 | return err; | |
2758 | ||
6b96018b | 2759 | if (copy_in_user(uifr32, uifr, sizeof(struct compat_ifreq))) |
7a229387 AB |
2760 | return -EFAULT; |
2761 | ||
2762 | return 0; | |
2763 | } | |
2764 | ||
6b96018b | 2765 | static int dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 2766 | { |
6b96018b | 2767 | struct compat_ifconf ifc32; |
7a229387 AB |
2768 | struct ifconf ifc; |
2769 | struct ifconf __user *uifc; | |
6b96018b | 2770 | struct compat_ifreq __user *ifr32; |
7a229387 AB |
2771 | struct ifreq __user *ifr; |
2772 | unsigned int i, j; | |
2773 | int err; | |
2774 | ||
6b96018b | 2775 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2776 | return -EFAULT; |
2777 | ||
43da5f2e | 2778 | memset(&ifc, 0, sizeof(ifc)); |
7a229387 AB |
2779 | if (ifc32.ifcbuf == 0) { |
2780 | ifc32.ifc_len = 0; | |
2781 | ifc.ifc_len = 0; | |
2782 | ifc.ifc_req = NULL; | |
2783 | uifc = compat_alloc_user_space(sizeof(struct ifconf)); | |
2784 | } else { | |
c6d409cf ED |
2785 | size_t len = ((ifc32.ifc_len / sizeof(struct compat_ifreq)) + 1) * |
2786 | sizeof(struct ifreq); | |
7a229387 AB |
2787 | uifc = compat_alloc_user_space(sizeof(struct ifconf) + len); |
2788 | ifc.ifc_len = len; | |
2789 | ifr = ifc.ifc_req = (void __user *)(uifc + 1); | |
2790 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
c6d409cf | 2791 | for (i = 0; i < ifc32.ifc_len; i += sizeof(struct compat_ifreq)) { |
6b96018b | 2792 | if (copy_in_user(ifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2793 | return -EFAULT; |
2794 | ifr++; | |
2795 | ifr32++; | |
2796 | } | |
2797 | } | |
2798 | if (copy_to_user(uifc, &ifc, sizeof(struct ifconf))) | |
2799 | return -EFAULT; | |
2800 | ||
6b96018b | 2801 | err = dev_ioctl(net, SIOCGIFCONF, uifc); |
7a229387 AB |
2802 | if (err) |
2803 | return err; | |
2804 | ||
2805 | if (copy_from_user(&ifc, uifc, sizeof(struct ifconf))) | |
2806 | return -EFAULT; | |
2807 | ||
2808 | ifr = ifc.ifc_req; | |
2809 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
2810 | for (i = 0, j = 0; | |
c6d409cf ED |
2811 | i + sizeof(struct compat_ifreq) <= ifc32.ifc_len && j < ifc.ifc_len; |
2812 | i += sizeof(struct compat_ifreq), j += sizeof(struct ifreq)) { | |
2813 | if (copy_in_user(ifr32, ifr, sizeof(struct compat_ifreq))) | |
7a229387 AB |
2814 | return -EFAULT; |
2815 | ifr32++; | |
2816 | ifr++; | |
2817 | } | |
2818 | ||
2819 | if (ifc32.ifcbuf == 0) { | |
2820 | /* Translate from 64-bit structure multiple to | |
2821 | * a 32-bit one. | |
2822 | */ | |
2823 | i = ifc.ifc_len; | |
6b96018b | 2824 | i = ((i / sizeof(struct ifreq)) * sizeof(struct compat_ifreq)); |
7a229387 AB |
2825 | ifc32.ifc_len = i; |
2826 | } else { | |
2827 | ifc32.ifc_len = i; | |
2828 | } | |
6b96018b | 2829 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2830 | return -EFAULT; |
2831 | ||
2832 | return 0; | |
2833 | } | |
2834 | ||
6b96018b | 2835 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 2836 | { |
3a7da39d BH |
2837 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
2838 | bool convert_in = false, convert_out = false; | |
2839 | size_t buf_size = ALIGN(sizeof(struct ifreq), 8); | |
2840 | struct ethtool_rxnfc __user *rxnfc; | |
7a229387 | 2841 | struct ifreq __user *ifr; |
3a7da39d BH |
2842 | u32 rule_cnt = 0, actual_rule_cnt; |
2843 | u32 ethcmd; | |
7a229387 | 2844 | u32 data; |
3a7da39d | 2845 | int ret; |
7a229387 | 2846 | |
3a7da39d BH |
2847 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
2848 | return -EFAULT; | |
7a229387 | 2849 | |
3a7da39d BH |
2850 | compat_rxnfc = compat_ptr(data); |
2851 | ||
2852 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
2853 | return -EFAULT; |
2854 | ||
3a7da39d BH |
2855 | /* Most ethtool structures are defined without padding. |
2856 | * Unfortunately struct ethtool_rxnfc is an exception. | |
2857 | */ | |
2858 | switch (ethcmd) { | |
2859 | default: | |
2860 | break; | |
2861 | case ETHTOOL_GRXCLSRLALL: | |
2862 | /* Buffer size is variable */ | |
2863 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
2864 | return -EFAULT; | |
2865 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
2866 | return -ENOMEM; | |
2867 | buf_size += rule_cnt * sizeof(u32); | |
2868 | /* fall through */ | |
2869 | case ETHTOOL_GRXRINGS: | |
2870 | case ETHTOOL_GRXCLSRLCNT: | |
2871 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 2872 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
2873 | convert_out = true; |
2874 | /* fall through */ | |
2875 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
2876 | buf_size += sizeof(struct ethtool_rxnfc); |
2877 | convert_in = true; | |
2878 | break; | |
2879 | } | |
2880 | ||
2881 | ifr = compat_alloc_user_space(buf_size); | |
954b1244 | 2882 | rxnfc = (void __user *)ifr + ALIGN(sizeof(struct ifreq), 8); |
3a7da39d BH |
2883 | |
2884 | if (copy_in_user(&ifr->ifr_name, &ifr32->ifr_name, IFNAMSIZ)) | |
7a229387 AB |
2885 | return -EFAULT; |
2886 | ||
3a7da39d BH |
2887 | if (put_user(convert_in ? rxnfc : compat_ptr(data), |
2888 | &ifr->ifr_ifru.ifru_data)) | |
7a229387 AB |
2889 | return -EFAULT; |
2890 | ||
3a7da39d | 2891 | if (convert_in) { |
127fe533 | 2892 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
2893 | * fs.ring_cookie and at the end of fs, but nowhere else. |
2894 | */ | |
127fe533 AD |
2895 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
2896 | sizeof(compat_rxnfc->fs.m_ext) != | |
2897 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
2898 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
2899 | BUILD_BUG_ON( |
2900 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
2901 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
2902 | offsetof(struct ethtool_rxnfc, fs.location) - | |
2903 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
2904 | ||
2905 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
2906 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
2907 | (void __user *)rxnfc) || | |
3a7da39d BH |
2908 | copy_in_user(&rxnfc->fs.ring_cookie, |
2909 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 SH |
2910 | (void __user *)(&rxnfc->fs.location + 1) - |
2911 | (void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2912 | copy_in_user(&rxnfc->rule_cnt, &compat_rxnfc->rule_cnt, |
2913 | sizeof(rxnfc->rule_cnt))) | |
2914 | return -EFAULT; | |
2915 | } | |
2916 | ||
2917 | ret = dev_ioctl(net, SIOCETHTOOL, ifr); | |
2918 | if (ret) | |
2919 | return ret; | |
2920 | ||
2921 | if (convert_out) { | |
2922 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
2923 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
2924 | (const void __user *)rxnfc) || | |
3a7da39d BH |
2925 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
2926 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
2927 | (const void __user *)(&rxnfc->fs.location + 1) - |
2928 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2929 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
2930 | sizeof(rxnfc->rule_cnt))) | |
2931 | return -EFAULT; | |
2932 | ||
2933 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
2934 | /* As an optimisation, we only copy the actual | |
2935 | * number of rules that the underlying | |
2936 | * function returned. Since Mallory might | |
2937 | * change the rule count in user memory, we | |
2938 | * check that it is less than the rule count | |
2939 | * originally given (as the user buffer size), | |
2940 | * which has been range-checked. | |
2941 | */ | |
2942 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
2943 | return -EFAULT; | |
2944 | if (actual_rule_cnt < rule_cnt) | |
2945 | rule_cnt = actual_rule_cnt; | |
2946 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
2947 | &rxnfc->rule_locs[0], | |
2948 | rule_cnt * sizeof(u32))) | |
2949 | return -EFAULT; | |
2950 | } | |
2951 | } | |
2952 | ||
2953 | return 0; | |
7a229387 AB |
2954 | } |
2955 | ||
7a50a240 AB |
2956 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
2957 | { | |
2958 | void __user *uptr; | |
2959 | compat_uptr_t uptr32; | |
2960 | struct ifreq __user *uifr; | |
2961 | ||
c6d409cf | 2962 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
7a50a240 AB |
2963 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
2964 | return -EFAULT; | |
2965 | ||
2966 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
2967 | return -EFAULT; | |
2968 | ||
2969 | uptr = compat_ptr(uptr32); | |
2970 | ||
2971 | if (put_user(uptr, &uifr->ifr_settings.ifs_ifsu.raw_hdlc)) | |
2972 | return -EFAULT; | |
2973 | ||
2974 | return dev_ioctl(net, SIOCWANDEV, uifr); | |
2975 | } | |
2976 | ||
6b96018b AB |
2977 | static int bond_ioctl(struct net *net, unsigned int cmd, |
2978 | struct compat_ifreq __user *ifr32) | |
7a229387 AB |
2979 | { |
2980 | struct ifreq kifr; | |
2981 | struct ifreq __user *uifr; | |
7a229387 AB |
2982 | mm_segment_t old_fs; |
2983 | int err; | |
2984 | u32 data; | |
2985 | void __user *datap; | |
2986 | ||
2987 | switch (cmd) { | |
2988 | case SIOCBONDENSLAVE: | |
2989 | case SIOCBONDRELEASE: | |
2990 | case SIOCBONDSETHWADDR: | |
2991 | case SIOCBONDCHANGEACTIVE: | |
6b96018b | 2992 | if (copy_from_user(&kifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2993 | return -EFAULT; |
2994 | ||
2995 | old_fs = get_fs(); | |
c6d409cf | 2996 | set_fs(KERNEL_DS); |
c3f52ae6 | 2997 | err = dev_ioctl(net, cmd, |
2998 | (struct ifreq __user __force *) &kifr); | |
c6d409cf | 2999 | set_fs(old_fs); |
7a229387 AB |
3000 | |
3001 | return err; | |
3002 | case SIOCBONDSLAVEINFOQUERY: | |
3003 | case SIOCBONDINFOQUERY: | |
3004 | uifr = compat_alloc_user_space(sizeof(*uifr)); | |
3005 | if (copy_in_user(&uifr->ifr_name, &ifr32->ifr_name, IFNAMSIZ)) | |
3006 | return -EFAULT; | |
3007 | ||
3008 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) | |
3009 | return -EFAULT; | |
3010 | ||
3011 | datap = compat_ptr(data); | |
3012 | if (put_user(datap, &uifr->ifr_ifru.ifru_data)) | |
3013 | return -EFAULT; | |
3014 | ||
6b96018b | 3015 | return dev_ioctl(net, cmd, uifr); |
7a229387 | 3016 | default: |
07d106d0 | 3017 | return -ENOIOCTLCMD; |
ccbd6a5a | 3018 | } |
7a229387 AB |
3019 | } |
3020 | ||
6b96018b AB |
3021 | static int siocdevprivate_ioctl(struct net *net, unsigned int cmd, |
3022 | struct compat_ifreq __user *u_ifreq32) | |
7a229387 AB |
3023 | { |
3024 | struct ifreq __user *u_ifreq64; | |
7a229387 AB |
3025 | char tmp_buf[IFNAMSIZ]; |
3026 | void __user *data64; | |
3027 | u32 data32; | |
3028 | ||
3029 | if (copy_from_user(&tmp_buf[0], &(u_ifreq32->ifr_ifrn.ifrn_name[0]), | |
3030 | IFNAMSIZ)) | |
3031 | return -EFAULT; | |
3032 | if (__get_user(data32, &u_ifreq32->ifr_ifru.ifru_data)) | |
3033 | return -EFAULT; | |
3034 | data64 = compat_ptr(data32); | |
3035 | ||
3036 | u_ifreq64 = compat_alloc_user_space(sizeof(*u_ifreq64)); | |
3037 | ||
3038 | /* Don't check these user accesses, just let that get trapped | |
3039 | * in the ioctl handler instead. | |
3040 | */ | |
3041 | if (copy_to_user(&u_ifreq64->ifr_ifrn.ifrn_name[0], &tmp_buf[0], | |
3042 | IFNAMSIZ)) | |
3043 | return -EFAULT; | |
3044 | if (__put_user(data64, &u_ifreq64->ifr_ifru.ifru_data)) | |
3045 | return -EFAULT; | |
3046 | ||
6b96018b | 3047 | return dev_ioctl(net, cmd, u_ifreq64); |
7a229387 AB |
3048 | } |
3049 | ||
6b96018b AB |
3050 | static int dev_ifsioc(struct net *net, struct socket *sock, |
3051 | unsigned int cmd, struct compat_ifreq __user *uifr32) | |
7a229387 | 3052 | { |
a2116ed2 | 3053 | struct ifreq __user *uifr; |
7a229387 AB |
3054 | int err; |
3055 | ||
a2116ed2 AB |
3056 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
3057 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
3058 | return -EFAULT; | |
3059 | ||
3060 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
3061 | ||
7a229387 AB |
3062 | if (!err) { |
3063 | switch (cmd) { | |
3064 | case SIOCGIFFLAGS: | |
3065 | case SIOCGIFMETRIC: | |
3066 | case SIOCGIFMTU: | |
3067 | case SIOCGIFMEM: | |
3068 | case SIOCGIFHWADDR: | |
3069 | case SIOCGIFINDEX: | |
3070 | case SIOCGIFADDR: | |
3071 | case SIOCGIFBRDADDR: | |
3072 | case SIOCGIFDSTADDR: | |
3073 | case SIOCGIFNETMASK: | |
fab2532b | 3074 | case SIOCGIFPFLAGS: |
7a229387 | 3075 | case SIOCGIFTXQLEN: |
fab2532b AB |
3076 | case SIOCGMIIPHY: |
3077 | case SIOCGMIIREG: | |
a2116ed2 | 3078 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
7a229387 AB |
3079 | err = -EFAULT; |
3080 | break; | |
3081 | } | |
3082 | } | |
3083 | return err; | |
3084 | } | |
3085 | ||
a2116ed2 AB |
3086 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
3087 | struct compat_ifreq __user *uifr32) | |
3088 | { | |
3089 | struct ifreq ifr; | |
3090 | struct compat_ifmap __user *uifmap32; | |
3091 | mm_segment_t old_fs; | |
3092 | int err; | |
3093 | ||
3094 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
3095 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3096 | err |= __get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); | |
3097 | err |= __get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3098 | err |= __get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3099 | err |= __get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3100 | err |= __get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3101 | err |= __get_user(ifr.ifr_map.port, &uifmap32->port); | |
3102 | if (err) | |
3103 | return -EFAULT; | |
3104 | ||
3105 | old_fs = get_fs(); | |
c6d409cf | 3106 | set_fs(KERNEL_DS); |
c3f52ae6 | 3107 | err = dev_ioctl(net, cmd, (void __user __force *)&ifr); |
c6d409cf | 3108 | set_fs(old_fs); |
a2116ed2 AB |
3109 | |
3110 | if (cmd == SIOCGIFMAP && !err) { | |
3111 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3112 | err |= __put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); | |
3113 | err |= __put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3114 | err |= __put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3115 | err |= __put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3116 | err |= __put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3117 | err |= __put_user(ifr.ifr_map.port, &uifmap32->port); | |
3118 | if (err) | |
3119 | err = -EFAULT; | |
3120 | } | |
3121 | return err; | |
3122 | } | |
3123 | ||
3124 | static int compat_siocshwtstamp(struct net *net, struct compat_ifreq __user *uifr32) | |
3125 | { | |
3126 | void __user *uptr; | |
3127 | compat_uptr_t uptr32; | |
3128 | struct ifreq __user *uifr; | |
3129 | ||
c6d409cf | 3130 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
a2116ed2 AB |
3131 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
3132 | return -EFAULT; | |
3133 | ||
3134 | if (get_user(uptr32, &uifr32->ifr_data)) | |
3135 | return -EFAULT; | |
3136 | ||
3137 | uptr = compat_ptr(uptr32); | |
3138 | ||
3139 | if (put_user(uptr, &uifr->ifr_data)) | |
3140 | return -EFAULT; | |
3141 | ||
3142 | return dev_ioctl(net, SIOCSHWTSTAMP, uifr); | |
3143 | } | |
3144 | ||
7a229387 | 3145 | struct rtentry32 { |
c6d409cf | 3146 | u32 rt_pad1; |
7a229387 AB |
3147 | struct sockaddr rt_dst; /* target address */ |
3148 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
3149 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
3150 | unsigned short rt_flags; |
3151 | short rt_pad2; | |
3152 | u32 rt_pad3; | |
3153 | unsigned char rt_tos; | |
3154 | unsigned char rt_class; | |
3155 | short rt_pad4; | |
3156 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 3157 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
3158 | u32 rt_mtu; /* per route MTU/Window */ |
3159 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
3160 | unsigned short rt_irtt; /* Initial RTT */ |
3161 | }; | |
3162 | ||
3163 | struct in6_rtmsg32 { | |
3164 | struct in6_addr rtmsg_dst; | |
3165 | struct in6_addr rtmsg_src; | |
3166 | struct in6_addr rtmsg_gateway; | |
3167 | u32 rtmsg_type; | |
3168 | u16 rtmsg_dst_len; | |
3169 | u16 rtmsg_src_len; | |
3170 | u32 rtmsg_metric; | |
3171 | u32 rtmsg_info; | |
3172 | u32 rtmsg_flags; | |
3173 | s32 rtmsg_ifindex; | |
3174 | }; | |
3175 | ||
6b96018b AB |
3176 | static int routing_ioctl(struct net *net, struct socket *sock, |
3177 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3178 | { |
3179 | int ret; | |
3180 | void *r = NULL; | |
3181 | struct in6_rtmsg r6; | |
3182 | struct rtentry r4; | |
3183 | char devname[16]; | |
3184 | u32 rtdev; | |
3185 | mm_segment_t old_fs = get_fs(); | |
3186 | ||
6b96018b AB |
3187 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3188 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3189 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3190 | 3 * sizeof(struct in6_addr)); |
c6d409cf ED |
3191 | ret |= __get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3192 | ret |= __get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3193 | ret |= __get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3194 | ret |= __get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3195 | ret |= __get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3196 | ret |= __get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3197 | ret |= __get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3198 | |
3199 | r = (void *) &r6; | |
3200 | } else { /* ipv4 */ | |
6b96018b | 3201 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3202 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3203 | 3 * sizeof(struct sockaddr)); |
c6d409cf ED |
3204 | ret |= __get_user(r4.rt_flags, &(ur4->rt_flags)); |
3205 | ret |= __get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3206 | ret |= __get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3207 | ret |= __get_user(r4.rt_window, &(ur4->rt_window)); | |
3208 | ret |= __get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3209 | ret |= __get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3210 | if (rtdev) { |
c6d409cf | 3211 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3212 | r4.rt_dev = (char __user __force *)devname; |
3213 | devname[15] = 0; | |
7a229387 AB |
3214 | } else |
3215 | r4.rt_dev = NULL; | |
3216 | ||
3217 | r = (void *) &r4; | |
3218 | } | |
3219 | ||
3220 | if (ret) { | |
3221 | ret = -EFAULT; | |
3222 | goto out; | |
3223 | } | |
3224 | ||
c6d409cf | 3225 | set_fs(KERNEL_DS); |
6b96018b | 3226 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3227 | set_fs(old_fs); |
7a229387 AB |
3228 | |
3229 | out: | |
7a229387 AB |
3230 | return ret; |
3231 | } | |
3232 | ||
3233 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3234 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3235 | * use compatible ioctls |
7a229387 | 3236 | */ |
6b96018b | 3237 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3238 | { |
6b96018b | 3239 | compat_ulong_t tmp; |
7a229387 | 3240 | |
6b96018b | 3241 | if (get_user(tmp, argp)) |
7a229387 AB |
3242 | return -EFAULT; |
3243 | if (tmp == BRCTL_GET_VERSION) | |
3244 | return BRCTL_VERSION + 1; | |
3245 | return -EINVAL; | |
3246 | } | |
3247 | ||
6b96018b AB |
3248 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3249 | unsigned int cmd, unsigned long arg) | |
3250 | { | |
3251 | void __user *argp = compat_ptr(arg); | |
3252 | struct sock *sk = sock->sk; | |
3253 | struct net *net = sock_net(sk); | |
7a229387 | 3254 | |
6b96018b AB |
3255 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
3256 | return siocdevprivate_ioctl(net, cmd, argp); | |
3257 | ||
3258 | switch (cmd) { | |
3259 | case SIOCSIFBR: | |
3260 | case SIOCGIFBR: | |
3261 | return old_bridge_ioctl(argp); | |
3262 | case SIOCGIFNAME: | |
3263 | return dev_ifname32(net, argp); | |
3264 | case SIOCGIFCONF: | |
3265 | return dev_ifconf(net, argp); | |
3266 | case SIOCETHTOOL: | |
3267 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3268 | case SIOCWANDEV: |
3269 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3270 | case SIOCGIFMAP: |
3271 | case SIOCSIFMAP: | |
3272 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3273 | case SIOCBONDENSLAVE: |
3274 | case SIOCBONDRELEASE: | |
3275 | case SIOCBONDSETHWADDR: | |
3276 | case SIOCBONDSLAVEINFOQUERY: | |
3277 | case SIOCBONDINFOQUERY: | |
3278 | case SIOCBONDCHANGEACTIVE: | |
3279 | return bond_ioctl(net, cmd, argp); | |
3280 | case SIOCADDRT: | |
3281 | case SIOCDELRT: | |
3282 | return routing_ioctl(net, sock, cmd, argp); | |
3283 | case SIOCGSTAMP: | |
3284 | return do_siocgstamp(net, sock, cmd, argp); | |
3285 | case SIOCGSTAMPNS: | |
3286 | return do_siocgstampns(net, sock, cmd, argp); | |
a2116ed2 AB |
3287 | case SIOCSHWTSTAMP: |
3288 | return compat_siocshwtstamp(net, argp); | |
6b96018b AB |
3289 | |
3290 | case FIOSETOWN: | |
3291 | case SIOCSPGRP: | |
3292 | case FIOGETOWN: | |
3293 | case SIOCGPGRP: | |
3294 | case SIOCBRADDBR: | |
3295 | case SIOCBRDELBR: | |
3296 | case SIOCGIFVLAN: | |
3297 | case SIOCSIFVLAN: | |
3298 | case SIOCADDDLCI: | |
3299 | case SIOCDELDLCI: | |
3c2a0909 | 3300 | case SIOCKILLADDR: |
6b96018b AB |
3301 | return sock_ioctl(file, cmd, arg); |
3302 | ||
3303 | case SIOCGIFFLAGS: | |
3304 | case SIOCSIFFLAGS: | |
3305 | case SIOCGIFMETRIC: | |
3306 | case SIOCSIFMETRIC: | |
3307 | case SIOCGIFMTU: | |
3308 | case SIOCSIFMTU: | |
3309 | case SIOCGIFMEM: | |
3310 | case SIOCSIFMEM: | |
3311 | case SIOCGIFHWADDR: | |
3312 | case SIOCSIFHWADDR: | |
3313 | case SIOCADDMULTI: | |
3314 | case SIOCDELMULTI: | |
3315 | case SIOCGIFINDEX: | |
6b96018b AB |
3316 | case SIOCGIFADDR: |
3317 | case SIOCSIFADDR: | |
3318 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3319 | case SIOCDIFADDR: |
6b96018b AB |
3320 | case SIOCGIFBRDADDR: |
3321 | case SIOCSIFBRDADDR: | |
3322 | case SIOCGIFDSTADDR: | |
3323 | case SIOCSIFDSTADDR: | |
3324 | case SIOCGIFNETMASK: | |
3325 | case SIOCSIFNETMASK: | |
3326 | case SIOCSIFPFLAGS: | |
3327 | case SIOCGIFPFLAGS: | |
3328 | case SIOCGIFTXQLEN: | |
3329 | case SIOCSIFTXQLEN: | |
3330 | case SIOCBRADDIF: | |
3331 | case SIOCBRDELIF: | |
9177efd3 AB |
3332 | case SIOCSIFNAME: |
3333 | case SIOCGMIIPHY: | |
3334 | case SIOCGMIIREG: | |
3335 | case SIOCSMIIREG: | |
6b96018b | 3336 | return dev_ifsioc(net, sock, cmd, argp); |
9177efd3 | 3337 | |
6b96018b AB |
3338 | case SIOCSARP: |
3339 | case SIOCGARP: | |
3340 | case SIOCDARP: | |
6b96018b | 3341 | case SIOCATMARK: |
9177efd3 AB |
3342 | return sock_do_ioctl(net, sock, cmd, arg); |
3343 | } | |
3344 | ||
6b96018b AB |
3345 | return -ENOIOCTLCMD; |
3346 | } | |
7a229387 | 3347 | |
95c96174 | 3348 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3349 | unsigned long arg) |
89bbfc95 SP |
3350 | { |
3351 | struct socket *sock = file->private_data; | |
3352 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3353 | struct sock *sk; |
3354 | struct net *net; | |
3355 | ||
3356 | sk = sock->sk; | |
3357 | net = sock_net(sk); | |
89bbfc95 SP |
3358 | |
3359 | if (sock->ops->compat_ioctl) | |
3360 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3361 | ||
87de87d5 DM |
3362 | if (ret == -ENOIOCTLCMD && |
3363 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3364 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3365 | ||
6b96018b AB |
3366 | if (ret == -ENOIOCTLCMD) |
3367 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3368 | ||
89bbfc95 SP |
3369 | return ret; |
3370 | } | |
3371 | #endif | |
3372 | ||
ac5a488e SS |
3373 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3374 | { | |
3375 | return sock->ops->bind(sock, addr, addrlen); | |
3376 | } | |
c6d409cf | 3377 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e SS |
3378 | |
3379 | int kernel_listen(struct socket *sock, int backlog) | |
3380 | { | |
3381 | return sock->ops->listen(sock, backlog); | |
3382 | } | |
c6d409cf | 3383 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e SS |
3384 | |
3385 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) | |
3386 | { | |
3387 | struct sock *sk = sock->sk; | |
3388 | int err; | |
3389 | ||
3390 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3391 | newsock); | |
3392 | if (err < 0) | |
3393 | goto done; | |
3394 | ||
3395 | err = sock->ops->accept(sock, *newsock, flags); | |
3396 | if (err < 0) { | |
3397 | sock_release(*newsock); | |
fa8705b0 | 3398 | *newsock = NULL; |
ac5a488e SS |
3399 | goto done; |
3400 | } | |
3401 | ||
3402 | (*newsock)->ops = sock->ops; | |
1b08534e | 3403 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3404 | |
3405 | done: | |
3406 | return err; | |
3407 | } | |
c6d409cf | 3408 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e SS |
3409 | |
3410 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, | |
4768fbcb | 3411 | int flags) |
ac5a488e SS |
3412 | { |
3413 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3414 | } | |
c6d409cf | 3415 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e SS |
3416 | |
3417 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr, | |
3418 | int *addrlen) | |
3419 | { | |
3420 | return sock->ops->getname(sock, addr, addrlen, 0); | |
3421 | } | |
c6d409cf | 3422 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e SS |
3423 | |
3424 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr, | |
3425 | int *addrlen) | |
3426 | { | |
3427 | return sock->ops->getname(sock, addr, addrlen, 1); | |
3428 | } | |
c6d409cf | 3429 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e SS |
3430 | |
3431 | int kernel_getsockopt(struct socket *sock, int level, int optname, | |
3432 | char *optval, int *optlen) | |
3433 | { | |
3434 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3435 | char __user *uoptval; |
3436 | int __user *uoptlen; | |
ac5a488e SS |
3437 | int err; |
3438 | ||
fb8621bb NK |
3439 | uoptval = (char __user __force *) optval; |
3440 | uoptlen = (int __user __force *) optlen; | |
3441 | ||
ac5a488e SS |
3442 | set_fs(KERNEL_DS); |
3443 | if (level == SOL_SOCKET) | |
fb8621bb | 3444 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3445 | else |
fb8621bb NK |
3446 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3447 | uoptlen); | |
ac5a488e SS |
3448 | set_fs(oldfs); |
3449 | return err; | |
3450 | } | |
c6d409cf | 3451 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e SS |
3452 | |
3453 | int kernel_setsockopt(struct socket *sock, int level, int optname, | |
b7058842 | 3454 | char *optval, unsigned int optlen) |
ac5a488e SS |
3455 | { |
3456 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3457 | char __user *uoptval; |
ac5a488e SS |
3458 | int err; |
3459 | ||
fb8621bb NK |
3460 | uoptval = (char __user __force *) optval; |
3461 | ||
ac5a488e SS |
3462 | set_fs(KERNEL_DS); |
3463 | if (level == SOL_SOCKET) | |
fb8621bb | 3464 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3465 | else |
fb8621bb | 3466 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3467 | optlen); |
3468 | set_fs(oldfs); | |
3469 | return err; | |
3470 | } | |
c6d409cf | 3471 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e SS |
3472 | |
3473 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, | |
3474 | size_t size, int flags) | |
3475 | { | |
3476 | if (sock->ops->sendpage) | |
3477 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3478 | ||
3479 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3480 | } | |
c6d409cf | 3481 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e SS |
3482 | |
3483 | int kernel_sock_ioctl(struct socket *sock, int cmd, unsigned long arg) | |
3484 | { | |
3485 | mm_segment_t oldfs = get_fs(); | |
3486 | int err; | |
3487 | ||
3488 | set_fs(KERNEL_DS); | |
3489 | err = sock->ops->ioctl(sock, cmd, arg); | |
3490 | set_fs(oldfs); | |
3491 | ||
3492 | return err; | |
3493 | } | |
c6d409cf | 3494 | EXPORT_SYMBOL(kernel_sock_ioctl); |
ac5a488e | 3495 | |
91cf45f0 TM |
3496 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3497 | { | |
3498 | return sock->ops->shutdown(sock, how); | |
3499 | } | |
91cf45f0 | 3500 | EXPORT_SYMBOL(kernel_sock_shutdown); |