]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * NET An implementation of the SOCKET network access protocol. | |
3 | * | |
4 | * Version: @(#)socket.c 1.1.93 18/02/95 | |
5 | * | |
6 | * Authors: Orest Zborowski, <obz@Kodak.COM> | |
02c30a84 | 7 | * Ross Biro |
1da177e4 LT |
8 | * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> |
9 | * | |
10 | * Fixes: | |
11 | * Anonymous : NOTSOCK/BADF cleanup. Error fix in | |
12 | * shutdown() | |
13 | * Alan Cox : verify_area() fixes | |
14 | * Alan Cox : Removed DDI | |
15 | * Jonathan Kamens : SOCK_DGRAM reconnect bug | |
16 | * Alan Cox : Moved a load of checks to the very | |
17 | * top level. | |
18 | * Alan Cox : Move address structures to/from user | |
19 | * mode above the protocol layers. | |
20 | * Rob Janssen : Allow 0 length sends. | |
21 | * Alan Cox : Asynchronous I/O support (cribbed from the | |
22 | * tty drivers). | |
23 | * Niibe Yutaka : Asynchronous I/O for writes (4.4BSD style) | |
24 | * Jeff Uphoff : Made max number of sockets command-line | |
25 | * configurable. | |
26 | * Matti Aarnio : Made the number of sockets dynamic, | |
27 | * to be allocated when needed, and mr. | |
28 | * Uphoff's max is used as max to be | |
29 | * allowed to allocate. | |
30 | * Linus : Argh. removed all the socket allocation | |
31 | * altogether: it's in the inode now. | |
32 | * Alan Cox : Made sock_alloc()/sock_release() public | |
33 | * for NetROM and future kernel nfsd type | |
34 | * stuff. | |
35 | * Alan Cox : sendmsg/recvmsg basics. | |
36 | * Tom Dyas : Export net symbols. | |
37 | * Marcin Dalecki : Fixed problems with CONFIG_NET="n". | |
38 | * Alan Cox : Added thread locking to sys_* calls | |
39 | * for sockets. May have errors at the | |
40 | * moment. | |
41 | * Kevin Buhr : Fixed the dumb errors in the above. | |
42 | * Andi Kleen : Some small cleanups, optimizations, | |
43 | * and fixed a copy_from_user() bug. | |
44 | * Tigran Aivazian : sys_send(args) calls sys_sendto(args, NULL, 0) | |
89bddce5 | 45 | * Tigran Aivazian : Made listen(2) backlog sanity checks |
1da177e4 LT |
46 | * protocol-independent |
47 | * | |
48 | * | |
49 | * This program is free software; you can redistribute it and/or | |
50 | * modify it under the terms of the GNU General Public License | |
51 | * as published by the Free Software Foundation; either version | |
52 | * 2 of the License, or (at your option) any later version. | |
53 | * | |
54 | * | |
55 | * This module is effectively the top level interface to the BSD socket | |
89bddce5 | 56 | * paradigm. |
1da177e4 LT |
57 | * |
58 | * Based upon Swansea University Computer Society NET3.039 | |
59 | */ | |
60 | ||
1da177e4 | 61 | #include <linux/mm.h> |
1da177e4 LT |
62 | #include <linux/socket.h> |
63 | #include <linux/file.h> | |
64 | #include <linux/net.h> | |
65 | #include <linux/interrupt.h> | |
aaca0bdc | 66 | #include <linux/thread_info.h> |
55737fda | 67 | #include <linux/rcupdate.h> |
1da177e4 LT |
68 | #include <linux/netdevice.h> |
69 | #include <linux/proc_fs.h> | |
70 | #include <linux/seq_file.h> | |
4a3e2f71 | 71 | #include <linux/mutex.h> |
1da177e4 | 72 | #include <linux/if_bridge.h> |
20380731 ACM |
73 | #include <linux/if_frad.h> |
74 | #include <linux/if_vlan.h> | |
408eccce | 75 | #include <linux/ptp_classify.h> |
1da177e4 LT |
76 | #include <linux/init.h> |
77 | #include <linux/poll.h> | |
78 | #include <linux/cache.h> | |
79 | #include <linux/module.h> | |
80 | #include <linux/highmem.h> | |
1da177e4 LT |
81 | #include <linux/mount.h> |
82 | #include <linux/security.h> | |
83 | #include <linux/syscalls.h> | |
84 | #include <linux/compat.h> | |
85 | #include <linux/kmod.h> | |
3ec3b2fb | 86 | #include <linux/audit.h> |
d86b5e0e | 87 | #include <linux/wireless.h> |
1b8d7ae4 | 88 | #include <linux/nsproxy.h> |
1fd7317d | 89 | #include <linux/magic.h> |
5a0e3ad6 | 90 | #include <linux/slab.h> |
600e1779 | 91 | #include <linux/xattr.h> |
1da177e4 LT |
92 | |
93 | #include <asm/uaccess.h> | |
94 | #include <asm/unistd.h> | |
95 | ||
96 | #include <net/compat.h> | |
87de87d5 | 97 | #include <net/wext.h> |
f8451725 | 98 | #include <net/cls_cgroup.h> |
1da177e4 LT |
99 | |
100 | #include <net/sock.h> | |
101 | #include <linux/netfilter.h> | |
102 | ||
6b96018b AB |
103 | #include <linux/if_tun.h> |
104 | #include <linux/ipv6_route.h> | |
105 | #include <linux/route.h> | |
6b96018b AB |
106 | #include <linux/sockios.h> |
107 | #include <linux/atalk.h> | |
076bb0c8 | 108 | #include <net/busy_poll.h> |
f24b9be5 | 109 | #include <linux/errqueue.h> |
06021292 | 110 | |
e0d1095a | 111 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
112 | unsigned int sysctl_net_busy_read __read_mostly; |
113 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 114 | #endif |
6b96018b | 115 | |
8ae5e030 AV |
116 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
117 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 118 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
119 | |
120 | static int sock_close(struct inode *inode, struct file *file); | |
121 | static unsigned int sock_poll(struct file *file, | |
122 | struct poll_table_struct *wait); | |
89bddce5 | 123 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
124 | #ifdef CONFIG_COMPAT |
125 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 126 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 127 | #endif |
1da177e4 | 128 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
129 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
130 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 131 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 132 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 133 | unsigned int flags); |
1da177e4 | 134 | |
1da177e4 LT |
135 | /* |
136 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
137 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
138 | */ | |
139 | ||
da7071d7 | 140 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
141 | .owner = THIS_MODULE, |
142 | .llseek = no_llseek, | |
8ae5e030 AV |
143 | .read_iter = sock_read_iter, |
144 | .write_iter = sock_write_iter, | |
1da177e4 LT |
145 | .poll = sock_poll, |
146 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
147 | #ifdef CONFIG_COMPAT |
148 | .compat_ioctl = compat_sock_ioctl, | |
149 | #endif | |
1da177e4 | 150 | .mmap = sock_mmap, |
1da177e4 LT |
151 | .release = sock_close, |
152 | .fasync = sock_fasync, | |
5274f052 JA |
153 | .sendpage = sock_sendpage, |
154 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 155 | .splice_read = sock_splice_read, |
1da177e4 LT |
156 | }; |
157 | ||
158 | /* | |
159 | * The protocol list. Each protocol is registered in here. | |
160 | */ | |
161 | ||
1da177e4 | 162 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 163 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 164 | |
1da177e4 LT |
165 | /* |
166 | * Statistics counters of the socket lists | |
167 | */ | |
168 | ||
c6d409cf | 169 | static DEFINE_PER_CPU(int, sockets_in_use); |
1da177e4 LT |
170 | |
171 | /* | |
89bddce5 SH |
172 | * Support routines. |
173 | * Move socket addresses back and forth across the kernel/user | |
174 | * divide and look after the messy bits. | |
1da177e4 LT |
175 | */ |
176 | ||
1da177e4 LT |
177 | /** |
178 | * move_addr_to_kernel - copy a socket address into kernel space | |
179 | * @uaddr: Address in user space | |
180 | * @kaddr: Address in kernel space | |
181 | * @ulen: Length in user space | |
182 | * | |
183 | * The address is copied into kernel space. If the provided address is | |
184 | * too long an error code of -EINVAL is returned. If the copy gives | |
185 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
186 | */ | |
187 | ||
43db362d | 188 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 189 | { |
230b1839 | 190 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 191 | return -EINVAL; |
89bddce5 | 192 | if (ulen == 0) |
1da177e4 | 193 | return 0; |
89bddce5 | 194 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 195 | return -EFAULT; |
3ec3b2fb | 196 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
197 | } |
198 | ||
199 | /** | |
200 | * move_addr_to_user - copy an address to user space | |
201 | * @kaddr: kernel space address | |
202 | * @klen: length of address in kernel | |
203 | * @uaddr: user space address | |
204 | * @ulen: pointer to user length field | |
205 | * | |
206 | * The value pointed to by ulen on entry is the buffer length available. | |
207 | * This is overwritten with the buffer space used. -EINVAL is returned | |
208 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
209 | * is returned if either the buffer or the length field are not | |
210 | * accessible. | |
211 | * After copying the data up to the limit the user specifies, the true | |
212 | * length of the data is written over the length limit the user | |
213 | * specified. Zero is returned for a success. | |
214 | */ | |
89bddce5 | 215 | |
43db362d | 216 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 217 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
218 | { |
219 | int err; | |
220 | int len; | |
221 | ||
68c6beb3 | 222 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
223 | err = get_user(len, ulen); |
224 | if (err) | |
1da177e4 | 225 | return err; |
89bddce5 SH |
226 | if (len > klen) |
227 | len = klen; | |
68c6beb3 | 228 | if (len < 0) |
1da177e4 | 229 | return -EINVAL; |
89bddce5 | 230 | if (len) { |
d6fe3945 SG |
231 | if (audit_sockaddr(klen, kaddr)) |
232 | return -ENOMEM; | |
89bddce5 | 233 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
234 | return -EFAULT; |
235 | } | |
236 | /* | |
89bddce5 SH |
237 | * "fromlen shall refer to the value before truncation.." |
238 | * 1003.1g | |
1da177e4 LT |
239 | */ |
240 | return __put_user(klen, ulen); | |
241 | } | |
242 | ||
e18b890b | 243 | static struct kmem_cache *sock_inode_cachep __read_mostly; |
1da177e4 LT |
244 | |
245 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
246 | { | |
247 | struct socket_alloc *ei; | |
eaefd110 | 248 | struct socket_wq *wq; |
89bddce5 | 249 | |
e94b1766 | 250 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
251 | if (!ei) |
252 | return NULL; | |
eaefd110 ED |
253 | wq = kmalloc(sizeof(*wq), GFP_KERNEL); |
254 | if (!wq) { | |
43815482 ED |
255 | kmem_cache_free(sock_inode_cachep, ei); |
256 | return NULL; | |
257 | } | |
eaefd110 ED |
258 | init_waitqueue_head(&wq->wait); |
259 | wq->fasync_list = NULL; | |
574aab1e | 260 | wq->flags = 0; |
eaefd110 | 261 | RCU_INIT_POINTER(ei->socket.wq, wq); |
89bddce5 | 262 | |
1da177e4 LT |
263 | ei->socket.state = SS_UNCONNECTED; |
264 | ei->socket.flags = 0; | |
265 | ei->socket.ops = NULL; | |
266 | ei->socket.sk = NULL; | |
267 | ei->socket.file = NULL; | |
1da177e4 LT |
268 | |
269 | return &ei->vfs_inode; | |
270 | } | |
271 | ||
272 | static void sock_destroy_inode(struct inode *inode) | |
273 | { | |
43815482 | 274 | struct socket_alloc *ei; |
eaefd110 | 275 | struct socket_wq *wq; |
43815482 ED |
276 | |
277 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
eaefd110 | 278 | wq = rcu_dereference_protected(ei->socket.wq, 1); |
61845220 | 279 | kfree_rcu(wq, rcu); |
43815482 | 280 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
281 | } |
282 | ||
51cc5068 | 283 | static void init_once(void *foo) |
1da177e4 | 284 | { |
89bddce5 | 285 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 286 | |
a35afb83 | 287 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 288 | } |
89bddce5 | 289 | |
1da177e4 LT |
290 | static int init_inodecache(void) |
291 | { | |
292 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
293 | sizeof(struct socket_alloc), |
294 | 0, | |
295 | (SLAB_HWCACHE_ALIGN | | |
296 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 297 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 298 | init_once); |
1da177e4 LT |
299 | if (sock_inode_cachep == NULL) |
300 | return -ENOMEM; | |
301 | return 0; | |
302 | } | |
303 | ||
b87221de | 304 | static const struct super_operations sockfs_ops = { |
c6d409cf ED |
305 | .alloc_inode = sock_alloc_inode, |
306 | .destroy_inode = sock_destroy_inode, | |
307 | .statfs = simple_statfs, | |
1da177e4 LT |
308 | }; |
309 | ||
c23fbb6b ED |
310 | /* |
311 | * sockfs_dname() is called from d_path(). | |
312 | */ | |
313 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
314 | { | |
315 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 316 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
317 | } |
318 | ||
3ba13d17 | 319 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 320 | .d_dname = sockfs_dname, |
1da177e4 LT |
321 | }; |
322 | ||
bba0bd31 AG |
323 | static int sockfs_xattr_get(const struct xattr_handler *handler, |
324 | struct dentry *dentry, struct inode *inode, | |
325 | const char *suffix, void *value, size_t size) | |
326 | { | |
327 | if (value) { | |
328 | if (dentry->d_name.len + 1 > size) | |
329 | return -ERANGE; | |
330 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
331 | } | |
332 | return dentry->d_name.len + 1; | |
333 | } | |
334 | ||
335 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" | |
336 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
337 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
338 | ||
339 | static const struct xattr_handler sockfs_xattr_handler = { | |
340 | .name = XATTR_NAME_SOCKPROTONAME, | |
341 | .get = sockfs_xattr_get, | |
342 | }; | |
343 | ||
4a590153 AG |
344 | static int sockfs_security_xattr_set(const struct xattr_handler *handler, |
345 | struct dentry *dentry, struct inode *inode, | |
346 | const char *suffix, const void *value, | |
347 | size_t size, int flags) | |
348 | { | |
349 | /* Handled by LSM. */ | |
350 | return -EAGAIN; | |
351 | } | |
352 | ||
353 | static const struct xattr_handler sockfs_security_xattr_handler = { | |
354 | .prefix = XATTR_SECURITY_PREFIX, | |
355 | .set = sockfs_security_xattr_set, | |
356 | }; | |
357 | ||
bba0bd31 AG |
358 | static const struct xattr_handler *sockfs_xattr_handlers[] = { |
359 | &sockfs_xattr_handler, | |
4a590153 | 360 | &sockfs_security_xattr_handler, |
bba0bd31 AG |
361 | NULL |
362 | }; | |
363 | ||
c74a1cbb AV |
364 | static struct dentry *sockfs_mount(struct file_system_type *fs_type, |
365 | int flags, const char *dev_name, void *data) | |
366 | { | |
bba0bd31 AG |
367 | return mount_pseudo_xattr(fs_type, "socket:", &sockfs_ops, |
368 | sockfs_xattr_handlers, | |
369 | &sockfs_dentry_operations, SOCKFS_MAGIC); | |
c74a1cbb AV |
370 | } |
371 | ||
372 | static struct vfsmount *sock_mnt __read_mostly; | |
373 | ||
374 | static struct file_system_type sock_fs_type = { | |
375 | .name = "sockfs", | |
376 | .mount = sockfs_mount, | |
377 | .kill_sb = kill_anon_super, | |
378 | }; | |
379 | ||
1da177e4 LT |
380 | /* |
381 | * Obtains the first available file descriptor and sets it up for use. | |
382 | * | |
39d8c1b6 DM |
383 | * These functions create file structures and maps them to fd space |
384 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
385 | * and file struct implicitly stored in sock->file. |
386 | * Note that another thread may close file descriptor before we return | |
387 | * from this function. We use the fact that now we do not refer | |
388 | * to socket after mapping. If one day we will need it, this | |
389 | * function will increment ref. count on file by 1. | |
390 | * | |
391 | * In any case returned fd MAY BE not valid! | |
392 | * This race condition is unavoidable | |
393 | * with shared fd spaces, we cannot solve it inside kernel, | |
394 | * but we take care of internal coherence yet. | |
395 | */ | |
396 | ||
aab174f0 | 397 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 398 | { |
7cbe66b6 | 399 | struct qstr name = { .name = "" }; |
2c48b9c4 | 400 | struct path path; |
7cbe66b6 | 401 | struct file *file; |
1da177e4 | 402 | |
600e1779 MY |
403 | if (dname) { |
404 | name.name = dname; | |
405 | name.len = strlen(name.name); | |
406 | } else if (sock->sk) { | |
407 | name.name = sock->sk->sk_prot_creator->name; | |
408 | name.len = strlen(name.name); | |
409 | } | |
4b936885 | 410 | path.dentry = d_alloc_pseudo(sock_mnt->mnt_sb, &name); |
28407630 AV |
411 | if (unlikely(!path.dentry)) |
412 | return ERR_PTR(-ENOMEM); | |
2c48b9c4 | 413 | path.mnt = mntget(sock_mnt); |
39d8c1b6 | 414 | |
2c48b9c4 | 415 | d_instantiate(path.dentry, SOCK_INODE(sock)); |
39d8c1b6 | 416 | |
2c48b9c4 | 417 | file = alloc_file(&path, FMODE_READ | FMODE_WRITE, |
ce8d2cdf | 418 | &socket_file_ops); |
b5ffe634 | 419 | if (IS_ERR(file)) { |
cc3808f8 | 420 | /* drop dentry, keep inode */ |
c5ef6035 | 421 | ihold(d_inode(path.dentry)); |
2c48b9c4 | 422 | path_put(&path); |
39b65252 | 423 | return file; |
cc3808f8 AV |
424 | } |
425 | ||
426 | sock->file = file; | |
77d27200 | 427 | file->f_flags = O_RDWR | (flags & O_NONBLOCK); |
39d8c1b6 | 428 | file->private_data = sock; |
28407630 | 429 | return file; |
39d8c1b6 | 430 | } |
56b31d1c | 431 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 432 | |
56b31d1c | 433 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
434 | { |
435 | struct file *newfile; | |
28407630 AV |
436 | int fd = get_unused_fd_flags(flags); |
437 | if (unlikely(fd < 0)) | |
438 | return fd; | |
39d8c1b6 | 439 | |
aab174f0 | 440 | newfile = sock_alloc_file(sock, flags, NULL); |
28407630 | 441 | if (likely(!IS_ERR(newfile))) { |
39d8c1b6 | 442 | fd_install(fd, newfile); |
28407630 AV |
443 | return fd; |
444 | } | |
7cbe66b6 | 445 | |
28407630 AV |
446 | put_unused_fd(fd); |
447 | return PTR_ERR(newfile); | |
1da177e4 LT |
448 | } |
449 | ||
406a3c63 | 450 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 451 | { |
6cb153ca BL |
452 | if (file->f_op == &socket_file_ops) |
453 | return file->private_data; /* set in sock_map_fd */ | |
454 | ||
23bb80d2 ED |
455 | *err = -ENOTSOCK; |
456 | return NULL; | |
6cb153ca | 457 | } |
406a3c63 | 458 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 459 | |
1da177e4 | 460 | /** |
c6d409cf | 461 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
462 | * @fd: file handle |
463 | * @err: pointer to an error code return | |
464 | * | |
465 | * The file handle passed in is locked and the socket it is bound | |
466 | * too is returned. If an error occurs the err pointer is overwritten | |
467 | * with a negative errno code and NULL is returned. The function checks | |
468 | * for both invalid handles and passing a handle which is not a socket. | |
469 | * | |
470 | * On a success the socket object pointer is returned. | |
471 | */ | |
472 | ||
473 | struct socket *sockfd_lookup(int fd, int *err) | |
474 | { | |
475 | struct file *file; | |
1da177e4 LT |
476 | struct socket *sock; |
477 | ||
89bddce5 SH |
478 | file = fget(fd); |
479 | if (!file) { | |
1da177e4 LT |
480 | *err = -EBADF; |
481 | return NULL; | |
482 | } | |
89bddce5 | 483 | |
6cb153ca BL |
484 | sock = sock_from_file(file, err); |
485 | if (!sock) | |
1da177e4 | 486 | fput(file); |
6cb153ca BL |
487 | return sock; |
488 | } | |
c6d409cf | 489 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 490 | |
6cb153ca BL |
491 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
492 | { | |
00e188ef | 493 | struct fd f = fdget(fd); |
6cb153ca BL |
494 | struct socket *sock; |
495 | ||
3672558c | 496 | *err = -EBADF; |
00e188ef AV |
497 | if (f.file) { |
498 | sock = sock_from_file(f.file, err); | |
499 | if (likely(sock)) { | |
500 | *fput_needed = f.flags; | |
6cb153ca | 501 | return sock; |
00e188ef AV |
502 | } |
503 | fdput(f); | |
1da177e4 | 504 | } |
6cb153ca | 505 | return NULL; |
1da177e4 LT |
506 | } |
507 | ||
600e1779 MY |
508 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
509 | size_t size) | |
510 | { | |
511 | ssize_t len; | |
512 | ssize_t used = 0; | |
513 | ||
c5ef6035 | 514 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
515 | if (len < 0) |
516 | return len; | |
517 | used += len; | |
518 | if (buffer) { | |
519 | if (size < used) | |
520 | return -ERANGE; | |
521 | buffer += len; | |
522 | } | |
523 | ||
524 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
525 | used += len; | |
526 | if (buffer) { | |
527 | if (size < used) | |
528 | return -ERANGE; | |
529 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
530 | buffer += len; | |
531 | } | |
532 | ||
533 | return used; | |
534 | } | |
535 | ||
86741ec2 LC |
536 | int sockfs_setattr(struct dentry *dentry, struct iattr *iattr) |
537 | { | |
538 | int err = simple_setattr(dentry, iattr); | |
539 | ||
540 | if (!err) { | |
541 | struct socket *sock = SOCKET_I(d_inode(dentry)); | |
542 | ||
543 | sock->sk->sk_uid = iattr->ia_uid; | |
544 | } | |
545 | ||
546 | return err; | |
547 | } | |
548 | ||
600e1779 | 549 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 550 | .listxattr = sockfs_listxattr, |
86741ec2 | 551 | .setattr = sockfs_setattr, |
600e1779 MY |
552 | }; |
553 | ||
1da177e4 LT |
554 | /** |
555 | * sock_alloc - allocate a socket | |
89bddce5 | 556 | * |
1da177e4 LT |
557 | * Allocate a new inode and socket object. The two are bound together |
558 | * and initialised. The socket is then returned. If we are out of inodes | |
559 | * NULL is returned. | |
560 | */ | |
561 | ||
f4a00aac | 562 | struct socket *sock_alloc(void) |
1da177e4 | 563 | { |
89bddce5 SH |
564 | struct inode *inode; |
565 | struct socket *sock; | |
1da177e4 | 566 | |
a209dfc7 | 567 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
568 | if (!inode) |
569 | return NULL; | |
570 | ||
571 | sock = SOCKET_I(inode); | |
572 | ||
29a020d3 | 573 | kmemcheck_annotate_bitfield(sock, type); |
85fe4025 | 574 | inode->i_ino = get_next_ino(); |
89bddce5 | 575 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
576 | inode->i_uid = current_fsuid(); |
577 | inode->i_gid = current_fsgid(); | |
600e1779 | 578 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 579 | |
19e8d69c | 580 | this_cpu_add(sockets_in_use, 1); |
1da177e4 LT |
581 | return sock; |
582 | } | |
f4a00aac | 583 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 584 | |
1da177e4 LT |
585 | /** |
586 | * sock_release - close a socket | |
587 | * @sock: socket to close | |
588 | * | |
589 | * The socket is released from the protocol stack if it has a release | |
590 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 591 | * an inode not a file. |
1da177e4 | 592 | */ |
89bddce5 | 593 | |
1da177e4 LT |
594 | void sock_release(struct socket *sock) |
595 | { | |
596 | if (sock->ops) { | |
597 | struct module *owner = sock->ops->owner; | |
598 | ||
599 | sock->ops->release(sock); | |
600 | sock->ops = NULL; | |
601 | module_put(owner); | |
602 | } | |
603 | ||
eaefd110 | 604 | if (rcu_dereference_protected(sock->wq, 1)->fasync_list) |
3410f22e | 605 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 606 | |
19e8d69c | 607 | this_cpu_sub(sockets_in_use, 1); |
1da177e4 LT |
608 | if (!sock->file) { |
609 | iput(SOCK_INODE(sock)); | |
610 | return; | |
611 | } | |
89bddce5 | 612 | sock->file = NULL; |
1da177e4 | 613 | } |
c6d409cf | 614 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 615 | |
c14ac945 | 616 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 617 | { |
140c55d4 ED |
618 | u8 flags = *tx_flags; |
619 | ||
c14ac945 | 620 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
621 | flags |= SKBTX_HW_TSTAMP; |
622 | ||
c14ac945 | 623 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
624 | flags |= SKBTX_SW_TSTAMP; |
625 | ||
c14ac945 | 626 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
627 | flags |= SKBTX_SCHED_TSTAMP; |
628 | ||
140c55d4 | 629 | *tx_flags = flags; |
20d49473 | 630 | } |
67cc0d40 | 631 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 632 | |
d8725c86 | 633 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 634 | { |
01e97e65 | 635 | int ret = sock->ops->sendmsg(sock, msg, msg_data_left(msg)); |
d8725c86 AV |
636 | BUG_ON(ret == -EIOCBQUEUED); |
637 | return ret; | |
1da177e4 LT |
638 | } |
639 | ||
d8725c86 | 640 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 641 | { |
d8725c86 | 642 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 643 | msg_data_left(msg)); |
228e548e | 644 | |
d8725c86 | 645 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 646 | } |
c6d409cf | 647 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 LT |
648 | |
649 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, | |
650 | struct kvec *vec, size_t num, size_t size) | |
651 | { | |
6aa24814 | 652 | iov_iter_kvec(&msg->msg_iter, WRITE | ITER_KVEC, vec, num, size); |
d8725c86 | 653 | return sock_sendmsg(sock, msg); |
1da177e4 | 654 | } |
c6d409cf | 655 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 656 | |
92f37fd2 ED |
657 | /* |
658 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
659 | */ | |
660 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
661 | struct sk_buff *skb) | |
662 | { | |
20d49473 | 663 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
f24b9be5 | 664 | struct scm_timestamping tss; |
20d49473 PO |
665 | int empty = 1; |
666 | struct skb_shared_hwtstamps *shhwtstamps = | |
667 | skb_hwtstamps(skb); | |
668 | ||
669 | /* Race occurred between timestamp enabling and packet | |
670 | receiving. Fill in the current time for now. */ | |
671 | if (need_software_tstamp && skb->tstamp.tv64 == 0) | |
672 | __net_timestamp(skb); | |
673 | ||
674 | if (need_software_tstamp) { | |
675 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
676 | struct timeval tv; | |
677 | skb_get_timestamp(skb, &tv); | |
678 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMP, | |
679 | sizeof(tv), &tv); | |
680 | } else { | |
f24b9be5 WB |
681 | struct timespec ts; |
682 | skb_get_timestampns(skb, &ts); | |
20d49473 | 683 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPNS, |
f24b9be5 | 684 | sizeof(ts), &ts); |
20d49473 PO |
685 | } |
686 | } | |
687 | ||
f24b9be5 | 688 | memset(&tss, 0, sizeof(tss)); |
c199105d | 689 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
f24b9be5 | 690 | ktime_to_timespec_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 691 | empty = 0; |
4d276eb6 | 692 | if (shhwtstamps && |
b9f40e21 | 693 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
f24b9be5 | 694 | ktime_to_timespec_cond(shhwtstamps->hwtstamp, tss.ts + 2)) |
4d276eb6 | 695 | empty = 0; |
20d49473 PO |
696 | if (!empty) |
697 | put_cmsg(msg, SOL_SOCKET, | |
f24b9be5 | 698 | SCM_TIMESTAMPING, sizeof(tss), &tss); |
92f37fd2 | 699 | } |
7c81fd8b ACM |
700 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
701 | ||
6e3e939f JB |
702 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
703 | struct sk_buff *skb) | |
704 | { | |
705 | int ack; | |
706 | ||
707 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
708 | return; | |
709 | if (!skb->wifi_acked_valid) | |
710 | return; | |
711 | ||
712 | ack = skb->wifi_acked; | |
713 | ||
714 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
715 | } | |
716 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
717 | ||
11165f14 | 718 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
719 | struct sk_buff *skb) | |
3b885787 | 720 | { |
744d5a3e | 721 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 722 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 723 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
724 | } |
725 | ||
767dd033 | 726 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
727 | struct sk_buff *skb) |
728 | { | |
729 | sock_recv_timestamp(msg, sk, skb); | |
730 | sock_recv_drops(msg, sk, skb); | |
731 | } | |
767dd033 | 732 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 733 | |
1b784140 | 734 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 735 | int flags) |
1da177e4 | 736 | { |
2da62906 | 737 | return sock->ops->recvmsg(sock, msg, msg_data_left(msg), flags); |
1da177e4 LT |
738 | } |
739 | ||
2da62906 | 740 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 741 | { |
2da62906 | 742 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 743 | |
2da62906 | 744 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 745 | } |
c6d409cf | 746 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 747 | |
c1249c0a ML |
748 | /** |
749 | * kernel_recvmsg - Receive a message from a socket (kernel space) | |
750 | * @sock: The socket to receive the message from | |
751 | * @msg: Received message | |
752 | * @vec: Input s/g array for message data | |
753 | * @num: Size of input s/g array | |
754 | * @size: Number of bytes to read | |
755 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
756 | * | |
757 | * On return the msg structure contains the scatter/gather array passed in the | |
758 | * vec argument. The array is modified so that it consists of the unfilled | |
759 | * portion of the original array. | |
760 | * | |
761 | * The returned value is the total number of bytes received, or an error. | |
762 | */ | |
89bddce5 SH |
763 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
764 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
765 | { |
766 | mm_segment_t oldfs = get_fs(); | |
767 | int result; | |
768 | ||
6aa24814 | 769 | iov_iter_kvec(&msg->msg_iter, READ | ITER_KVEC, vec, num, size); |
1da177e4 | 770 | set_fs(KERNEL_DS); |
2da62906 | 771 | result = sock_recvmsg(sock, msg, flags); |
1da177e4 LT |
772 | set_fs(oldfs); |
773 | return result; | |
774 | } | |
c6d409cf | 775 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 776 | |
ce1d4d3e CH |
777 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
778 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 779 | { |
1da177e4 LT |
780 | struct socket *sock; |
781 | int flags; | |
782 | ||
ce1d4d3e CH |
783 | sock = file->private_data; |
784 | ||
35f9c09f ED |
785 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
786 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
787 | flags |= more; | |
ce1d4d3e | 788 | |
e6949583 | 789 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 790 | } |
1da177e4 | 791 | |
9c55e01c | 792 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 793 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
794 | unsigned int flags) |
795 | { | |
796 | struct socket *sock = file->private_data; | |
797 | ||
997b37da RDC |
798 | if (unlikely(!sock->ops->splice_read)) |
799 | return -EINVAL; | |
800 | ||
9c55e01c JA |
801 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
802 | } | |
803 | ||
8ae5e030 | 804 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 805 | { |
6d652330 AV |
806 | struct file *file = iocb->ki_filp; |
807 | struct socket *sock = file->private_data; | |
0345f931 | 808 | struct msghdr msg = {.msg_iter = *to, |
809 | .msg_iocb = iocb}; | |
8ae5e030 | 810 | ssize_t res; |
ce1d4d3e | 811 | |
8ae5e030 AV |
812 | if (file->f_flags & O_NONBLOCK) |
813 | msg.msg_flags = MSG_DONTWAIT; | |
814 | ||
815 | if (iocb->ki_pos != 0) | |
1da177e4 | 816 | return -ESPIPE; |
027445c3 | 817 | |
66ee59af | 818 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
819 | return 0; |
820 | ||
2da62906 | 821 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
822 | *to = msg.msg_iter; |
823 | return res; | |
1da177e4 LT |
824 | } |
825 | ||
8ae5e030 | 826 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 827 | { |
6d652330 AV |
828 | struct file *file = iocb->ki_filp; |
829 | struct socket *sock = file->private_data; | |
0345f931 | 830 | struct msghdr msg = {.msg_iter = *from, |
831 | .msg_iocb = iocb}; | |
8ae5e030 | 832 | ssize_t res; |
1da177e4 | 833 | |
8ae5e030 | 834 | if (iocb->ki_pos != 0) |
ce1d4d3e | 835 | return -ESPIPE; |
027445c3 | 836 | |
8ae5e030 AV |
837 | if (file->f_flags & O_NONBLOCK) |
838 | msg.msg_flags = MSG_DONTWAIT; | |
839 | ||
6d652330 AV |
840 | if (sock->type == SOCK_SEQPACKET) |
841 | msg.msg_flags |= MSG_EOR; | |
842 | ||
d8725c86 | 843 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
844 | *from = msg.msg_iter; |
845 | return res; | |
1da177e4 LT |
846 | } |
847 | ||
1da177e4 LT |
848 | /* |
849 | * Atomic setting of ioctl hooks to avoid race | |
850 | * with module unload. | |
851 | */ | |
852 | ||
4a3e2f71 | 853 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 854 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 855 | |
881d966b | 856 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 857 | { |
4a3e2f71 | 858 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 859 | br_ioctl_hook = hook; |
4a3e2f71 | 860 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
861 | } |
862 | EXPORT_SYMBOL(brioctl_set); | |
863 | ||
4a3e2f71 | 864 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 865 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 866 | |
881d966b | 867 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 868 | { |
4a3e2f71 | 869 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 870 | vlan_ioctl_hook = hook; |
4a3e2f71 | 871 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
872 | } |
873 | EXPORT_SYMBOL(vlan_ioctl_set); | |
874 | ||
4a3e2f71 | 875 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 876 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 877 | |
89bddce5 | 878 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 879 | { |
4a3e2f71 | 880 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 881 | dlci_ioctl_hook = hook; |
4a3e2f71 | 882 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
883 | } |
884 | EXPORT_SYMBOL(dlci_ioctl_set); | |
885 | ||
6b96018b AB |
886 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
887 | unsigned int cmd, unsigned long arg) | |
888 | { | |
889 | int err; | |
890 | void __user *argp = (void __user *)arg; | |
891 | ||
892 | err = sock->ops->ioctl(sock, cmd, arg); | |
893 | ||
894 | /* | |
895 | * If this ioctl is unknown try to hand it down | |
896 | * to the NIC driver. | |
897 | */ | |
898 | if (err == -ENOIOCTLCMD) | |
899 | err = dev_ioctl(net, cmd, argp); | |
900 | ||
901 | return err; | |
902 | } | |
903 | ||
1da177e4 LT |
904 | /* |
905 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
906 | * what to do with it - that's up to the protocol still. | |
907 | */ | |
908 | ||
c62cce2c AV |
909 | static struct ns_common *get_net_ns(struct ns_common *ns) |
910 | { | |
911 | return &get_net(container_of(ns, struct net, ns))->ns; | |
912 | } | |
913 | ||
1da177e4 LT |
914 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
915 | { | |
916 | struct socket *sock; | |
881d966b | 917 | struct sock *sk; |
1da177e4 LT |
918 | void __user *argp = (void __user *)arg; |
919 | int pid, err; | |
881d966b | 920 | struct net *net; |
1da177e4 | 921 | |
b69aee04 | 922 | sock = file->private_data; |
881d966b | 923 | sk = sock->sk; |
3b1e0a65 | 924 | net = sock_net(sk); |
1da177e4 | 925 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) { |
881d966b | 926 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 927 | } else |
3d23e349 | 928 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 929 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
881d966b | 930 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 931 | } else |
3d23e349 | 932 | #endif |
89bddce5 | 933 | switch (cmd) { |
1da177e4 LT |
934 | case FIOSETOWN: |
935 | case SIOCSPGRP: | |
936 | err = -EFAULT; | |
937 | if (get_user(pid, (int __user *)argp)) | |
938 | break; | |
e0b93edd JL |
939 | f_setown(sock->file, pid, 1); |
940 | err = 0; | |
1da177e4 LT |
941 | break; |
942 | case FIOGETOWN: | |
943 | case SIOCGPGRP: | |
609d7fa9 | 944 | err = put_user(f_getown(sock->file), |
89bddce5 | 945 | (int __user *)argp); |
1da177e4 LT |
946 | break; |
947 | case SIOCGIFBR: | |
948 | case SIOCSIFBR: | |
949 | case SIOCBRADDBR: | |
950 | case SIOCBRDELBR: | |
951 | err = -ENOPKG; | |
952 | if (!br_ioctl_hook) | |
953 | request_module("bridge"); | |
954 | ||
4a3e2f71 | 955 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 956 | if (br_ioctl_hook) |
881d966b | 957 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 958 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
959 | break; |
960 | case SIOCGIFVLAN: | |
961 | case SIOCSIFVLAN: | |
962 | err = -ENOPKG; | |
963 | if (!vlan_ioctl_hook) | |
964 | request_module("8021q"); | |
965 | ||
4a3e2f71 | 966 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 967 | if (vlan_ioctl_hook) |
881d966b | 968 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 969 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 970 | break; |
1da177e4 LT |
971 | case SIOCADDDLCI: |
972 | case SIOCDELDLCI: | |
973 | err = -ENOPKG; | |
974 | if (!dlci_ioctl_hook) | |
975 | request_module("dlci"); | |
976 | ||
7512cbf6 PE |
977 | mutex_lock(&dlci_ioctl_mutex); |
978 | if (dlci_ioctl_hook) | |
1da177e4 | 979 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 980 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 | 981 | break; |
c62cce2c AV |
982 | case SIOCGSKNS: |
983 | err = -EPERM; | |
984 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
985 | break; | |
986 | ||
987 | err = open_related_ns(&net->ns, get_net_ns); | |
988 | break; | |
1da177e4 | 989 | default: |
6b96018b | 990 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 991 | break; |
89bddce5 | 992 | } |
1da177e4 LT |
993 | return err; |
994 | } | |
995 | ||
996 | int sock_create_lite(int family, int type, int protocol, struct socket **res) | |
997 | { | |
998 | int err; | |
999 | struct socket *sock = NULL; | |
89bddce5 | 1000 | |
1da177e4 LT |
1001 | err = security_socket_create(family, type, protocol, 1); |
1002 | if (err) | |
1003 | goto out; | |
1004 | ||
1005 | sock = sock_alloc(); | |
1006 | if (!sock) { | |
1007 | err = -ENOMEM; | |
1008 | goto out; | |
1009 | } | |
1010 | ||
1da177e4 | 1011 | sock->type = type; |
7420ed23 VY |
1012 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1013 | if (err) | |
1014 | goto out_release; | |
1015 | ||
1da177e4 LT |
1016 | out: |
1017 | *res = sock; | |
1018 | return err; | |
7420ed23 VY |
1019 | out_release: |
1020 | sock_release(sock); | |
1021 | sock = NULL; | |
1022 | goto out; | |
1da177e4 | 1023 | } |
c6d409cf | 1024 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1025 | |
1026 | /* No kernel lock held - perfect */ | |
89bddce5 | 1027 | static unsigned int sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1028 | { |
cbf55001 | 1029 | unsigned int busy_flag = 0; |
1da177e4 LT |
1030 | struct socket *sock; |
1031 | ||
1032 | /* | |
89bddce5 | 1033 | * We can't return errors to poll, so it's either yes or no. |
1da177e4 | 1034 | */ |
b69aee04 | 1035 | sock = file->private_data; |
2d48d67f | 1036 | |
cbf55001 | 1037 | if (sk_can_busy_loop(sock->sk)) { |
2d48d67f | 1038 | /* this socket can poll_ll so tell the system call */ |
cbf55001 | 1039 | busy_flag = POLL_BUSY_LOOP; |
2d48d67f ET |
1040 | |
1041 | /* once, only if requested by syscall */ | |
cbf55001 ET |
1042 | if (wait && (wait->_key & POLL_BUSY_LOOP)) |
1043 | sk_busy_loop(sock->sk, 1); | |
2d48d67f ET |
1044 | } |
1045 | ||
cbf55001 | 1046 | return busy_flag | sock->ops->poll(file, sock, wait); |
1da177e4 LT |
1047 | } |
1048 | ||
89bddce5 | 1049 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1050 | { |
b69aee04 | 1051 | struct socket *sock = file->private_data; |
1da177e4 LT |
1052 | |
1053 | return sock->ops->mmap(file, sock, vma); | |
1054 | } | |
1055 | ||
20380731 | 1056 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1057 | { |
1da177e4 LT |
1058 | sock_release(SOCKET_I(inode)); |
1059 | return 0; | |
1060 | } | |
1061 | ||
1062 | /* | |
1063 | * Update the socket async list | |
1064 | * | |
1065 | * Fasync_list locking strategy. | |
1066 | * | |
1067 | * 1. fasync_list is modified only under process context socket lock | |
1068 | * i.e. under semaphore. | |
1069 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1070 | * or under socket lock |
1da177e4 LT |
1071 | */ |
1072 | ||
1073 | static int sock_fasync(int fd, struct file *filp, int on) | |
1074 | { | |
989a2979 ED |
1075 | struct socket *sock = filp->private_data; |
1076 | struct sock *sk = sock->sk; | |
eaefd110 | 1077 | struct socket_wq *wq; |
1da177e4 | 1078 | |
989a2979 | 1079 | if (sk == NULL) |
1da177e4 | 1080 | return -EINVAL; |
1da177e4 LT |
1081 | |
1082 | lock_sock(sk); | |
1e1d04e6 | 1083 | wq = rcu_dereference_protected(sock->wq, lockdep_sock_is_held(sk)); |
eaefd110 | 1084 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1085 | |
eaefd110 | 1086 | if (!wq->fasync_list) |
989a2979 ED |
1087 | sock_reset_flag(sk, SOCK_FASYNC); |
1088 | else | |
bcdce719 | 1089 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1090 | |
989a2979 | 1091 | release_sock(sk); |
1da177e4 LT |
1092 | return 0; |
1093 | } | |
1094 | ||
ceb5d58b | 1095 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1096 | |
ceb5d58b | 1097 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1098 | { |
ceb5d58b | 1099 | if (!wq || !wq->fasync_list) |
1da177e4 | 1100 | return -1; |
ceb5d58b | 1101 | |
89bddce5 | 1102 | switch (how) { |
8d8ad9d7 | 1103 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1104 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1105 | break; |
1106 | goto call_kill; | |
8d8ad9d7 | 1107 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1108 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 LT |
1109 | break; |
1110 | /* fall through */ | |
8d8ad9d7 | 1111 | case SOCK_WAKE_IO: |
89bddce5 | 1112 | call_kill: |
43815482 | 1113 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1114 | break; |
8d8ad9d7 | 1115 | case SOCK_WAKE_URG: |
43815482 | 1116 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1117 | } |
ceb5d58b | 1118 | |
1da177e4 LT |
1119 | return 0; |
1120 | } | |
c6d409cf | 1121 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1122 | |
721db93a | 1123 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1124 | struct socket **res, int kern) |
1da177e4 LT |
1125 | { |
1126 | int err; | |
1127 | struct socket *sock; | |
55737fda | 1128 | const struct net_proto_family *pf; |
1da177e4 LT |
1129 | |
1130 | /* | |
89bddce5 | 1131 | * Check protocol is in range |
1da177e4 LT |
1132 | */ |
1133 | if (family < 0 || family >= NPROTO) | |
1134 | return -EAFNOSUPPORT; | |
1135 | if (type < 0 || type >= SOCK_MAX) | |
1136 | return -EINVAL; | |
1137 | ||
1138 | /* Compatibility. | |
1139 | ||
1140 | This uglymoron is moved from INET layer to here to avoid | |
1141 | deadlock in module load. | |
1142 | */ | |
1143 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1144 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1145 | current->comm); | |
1da177e4 LT |
1146 | family = PF_PACKET; |
1147 | } | |
1148 | ||
1149 | err = security_socket_create(family, type, protocol, kern); | |
1150 | if (err) | |
1151 | return err; | |
89bddce5 | 1152 | |
55737fda SH |
1153 | /* |
1154 | * Allocate the socket and allow the family to set things up. if | |
1155 | * the protocol is 0, the family is instructed to select an appropriate | |
1156 | * default. | |
1157 | */ | |
1158 | sock = sock_alloc(); | |
1159 | if (!sock) { | |
e87cc472 | 1160 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1161 | return -ENFILE; /* Not exactly a match, but its the |
1162 | closest posix thing */ | |
1163 | } | |
1164 | ||
1165 | sock->type = type; | |
1166 | ||
95a5afca | 1167 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1168 | /* Attempt to load a protocol module if the find failed. |
1169 | * | |
1170 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1171 | * requested real, full-featured networking support upon configuration. |
1172 | * Otherwise module support will break! | |
1173 | */ | |
190683a9 | 1174 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1175 | request_module("net-pf-%d", family); |
1da177e4 LT |
1176 | #endif |
1177 | ||
55737fda SH |
1178 | rcu_read_lock(); |
1179 | pf = rcu_dereference(net_families[family]); | |
1180 | err = -EAFNOSUPPORT; | |
1181 | if (!pf) | |
1182 | goto out_release; | |
1da177e4 LT |
1183 | |
1184 | /* | |
1185 | * We will call the ->create function, that possibly is in a loadable | |
1186 | * module, so we have to bump that loadable module refcnt first. | |
1187 | */ | |
55737fda | 1188 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1189 | goto out_release; |
1190 | ||
55737fda SH |
1191 | /* Now protected by module ref count */ |
1192 | rcu_read_unlock(); | |
1193 | ||
3f378b68 | 1194 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1195 | if (err < 0) |
1da177e4 | 1196 | goto out_module_put; |
a79af59e | 1197 | |
1da177e4 LT |
1198 | /* |
1199 | * Now to bump the refcnt of the [loadable] module that owns this | |
1200 | * socket at sock_release time we decrement its refcnt. | |
1201 | */ | |
55737fda SH |
1202 | if (!try_module_get(sock->ops->owner)) |
1203 | goto out_module_busy; | |
1204 | ||
1da177e4 LT |
1205 | /* |
1206 | * Now that we're done with the ->create function, the [loadable] | |
1207 | * module can have its refcnt decremented | |
1208 | */ | |
55737fda | 1209 | module_put(pf->owner); |
7420ed23 VY |
1210 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1211 | if (err) | |
3b185525 | 1212 | goto out_sock_release; |
55737fda | 1213 | *res = sock; |
1da177e4 | 1214 | |
55737fda SH |
1215 | return 0; |
1216 | ||
1217 | out_module_busy: | |
1218 | err = -EAFNOSUPPORT; | |
1da177e4 | 1219 | out_module_put: |
55737fda SH |
1220 | sock->ops = NULL; |
1221 | module_put(pf->owner); | |
1222 | out_sock_release: | |
1da177e4 | 1223 | sock_release(sock); |
55737fda SH |
1224 | return err; |
1225 | ||
1226 | out_release: | |
1227 | rcu_read_unlock(); | |
1228 | goto out_sock_release; | |
1da177e4 | 1229 | } |
721db93a | 1230 | EXPORT_SYMBOL(__sock_create); |
1da177e4 LT |
1231 | |
1232 | int sock_create(int family, int type, int protocol, struct socket **res) | |
1233 | { | |
1b8d7ae4 | 1234 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1235 | } |
c6d409cf | 1236 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1237 | |
eeb1bd5c | 1238 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1239 | { |
eeb1bd5c | 1240 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1241 | } |
c6d409cf | 1242 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1243 | |
3e0fa65f | 1244 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1da177e4 LT |
1245 | { |
1246 | int retval; | |
1247 | struct socket *sock; | |
a677a039 UD |
1248 | int flags; |
1249 | ||
e38b36f3 UD |
1250 | /* Check the SOCK_* constants for consistency. */ |
1251 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1252 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1253 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1254 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1255 | ||
a677a039 | 1256 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1257 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1258 | return -EINVAL; |
1259 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1260 | |
aaca0bdc UD |
1261 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1262 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1263 | ||
1da177e4 LT |
1264 | retval = sock_create(family, type, protocol, &sock); |
1265 | if (retval < 0) | |
1266 | goto out; | |
1267 | ||
77d27200 | 1268 | retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1269 | if (retval < 0) |
1270 | goto out_release; | |
1271 | ||
1272 | out: | |
1273 | /* It may be already another descriptor 8) Not kernel problem. */ | |
1274 | return retval; | |
1275 | ||
1276 | out_release: | |
1277 | sock_release(sock); | |
1278 | return retval; | |
1279 | } | |
1280 | ||
1281 | /* | |
1282 | * Create a pair of connected sockets. | |
1283 | */ | |
1284 | ||
3e0fa65f HC |
1285 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1286 | int __user *, usockvec) | |
1da177e4 LT |
1287 | { |
1288 | struct socket *sock1, *sock2; | |
1289 | int fd1, fd2, err; | |
db349509 | 1290 | struct file *newfile1, *newfile2; |
a677a039 UD |
1291 | int flags; |
1292 | ||
1293 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1294 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1295 | return -EINVAL; |
1296 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1297 | |
aaca0bdc UD |
1298 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1299 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1300 | ||
1da177e4 LT |
1301 | /* |
1302 | * Obtain the first socket and check if the underlying protocol | |
1303 | * supports the socketpair call. | |
1304 | */ | |
1305 | ||
1306 | err = sock_create(family, type, protocol, &sock1); | |
1307 | if (err < 0) | |
1308 | goto out; | |
1309 | ||
1310 | err = sock_create(family, type, protocol, &sock2); | |
1311 | if (err < 0) | |
1312 | goto out_release_1; | |
1313 | ||
1314 | err = sock1->ops->socketpair(sock1, sock2); | |
89bddce5 | 1315 | if (err < 0) |
1da177e4 LT |
1316 | goto out_release_both; |
1317 | ||
28407630 | 1318 | fd1 = get_unused_fd_flags(flags); |
bf3c23d1 DM |
1319 | if (unlikely(fd1 < 0)) { |
1320 | err = fd1; | |
db349509 | 1321 | goto out_release_both; |
bf3c23d1 | 1322 | } |
d73aa286 | 1323 | |
28407630 | 1324 | fd2 = get_unused_fd_flags(flags); |
198de4d7 AV |
1325 | if (unlikely(fd2 < 0)) { |
1326 | err = fd2; | |
d73aa286 | 1327 | goto out_put_unused_1; |
28407630 AV |
1328 | } |
1329 | ||
aab174f0 | 1330 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1331 | if (IS_ERR(newfile1)) { |
28407630 | 1332 | err = PTR_ERR(newfile1); |
d73aa286 | 1333 | goto out_put_unused_both; |
28407630 AV |
1334 | } |
1335 | ||
aab174f0 | 1336 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1337 | if (IS_ERR(newfile2)) { |
1338 | err = PTR_ERR(newfile2); | |
d73aa286 | 1339 | goto out_fput_1; |
db349509 AV |
1340 | } |
1341 | ||
d73aa286 YD |
1342 | err = put_user(fd1, &usockvec[0]); |
1343 | if (err) | |
1344 | goto out_fput_both; | |
1345 | ||
1346 | err = put_user(fd2, &usockvec[1]); | |
1347 | if (err) | |
1348 | goto out_fput_both; | |
1349 | ||
157cf649 | 1350 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1351 | |
db349509 AV |
1352 | fd_install(fd1, newfile1); |
1353 | fd_install(fd2, newfile2); | |
1da177e4 LT |
1354 | /* fd1 and fd2 may be already another descriptors. |
1355 | * Not kernel problem. | |
1356 | */ | |
1357 | ||
d73aa286 | 1358 | return 0; |
1da177e4 | 1359 | |
d73aa286 YD |
1360 | out_fput_both: |
1361 | fput(newfile2); | |
1362 | fput(newfile1); | |
1363 | put_unused_fd(fd2); | |
1364 | put_unused_fd(fd1); | |
1365 | goto out; | |
1366 | ||
1367 | out_fput_1: | |
1368 | fput(newfile1); | |
1369 | put_unused_fd(fd2); | |
1370 | put_unused_fd(fd1); | |
1371 | sock_release(sock2); | |
1372 | goto out; | |
1da177e4 | 1373 | |
d73aa286 YD |
1374 | out_put_unused_both: |
1375 | put_unused_fd(fd2); | |
1376 | out_put_unused_1: | |
1377 | put_unused_fd(fd1); | |
1da177e4 | 1378 | out_release_both: |
89bddce5 | 1379 | sock_release(sock2); |
1da177e4 | 1380 | out_release_1: |
89bddce5 | 1381 | sock_release(sock1); |
1da177e4 LT |
1382 | out: |
1383 | return err; | |
1384 | } | |
1385 | ||
1da177e4 LT |
1386 | /* |
1387 | * Bind a name to a socket. Nothing much to do here since it's | |
1388 | * the protocol's responsibility to handle the local address. | |
1389 | * | |
1390 | * We move the socket address to kernel space before we call | |
1391 | * the protocol layer (having also checked the address is ok). | |
1392 | */ | |
1393 | ||
20f37034 | 1394 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1da177e4 LT |
1395 | { |
1396 | struct socket *sock; | |
230b1839 | 1397 | struct sockaddr_storage address; |
6cb153ca | 1398 | int err, fput_needed; |
1da177e4 | 1399 | |
89bddce5 | 1400 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1401 | if (sock) { |
43db362d | 1402 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
89bddce5 SH |
1403 | if (err >= 0) { |
1404 | err = security_socket_bind(sock, | |
230b1839 | 1405 | (struct sockaddr *)&address, |
89bddce5 | 1406 | addrlen); |
6cb153ca BL |
1407 | if (!err) |
1408 | err = sock->ops->bind(sock, | |
89bddce5 | 1409 | (struct sockaddr *) |
230b1839 | 1410 | &address, addrlen); |
1da177e4 | 1411 | } |
6cb153ca | 1412 | fput_light(sock->file, fput_needed); |
89bddce5 | 1413 | } |
1da177e4 LT |
1414 | return err; |
1415 | } | |
1416 | ||
1da177e4 LT |
1417 | /* |
1418 | * Perform a listen. Basically, we allow the protocol to do anything | |
1419 | * necessary for a listen, and if that works, we mark the socket as | |
1420 | * ready for listening. | |
1421 | */ | |
1422 | ||
3e0fa65f | 1423 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1da177e4 LT |
1424 | { |
1425 | struct socket *sock; | |
6cb153ca | 1426 | int err, fput_needed; |
b8e1f9b5 | 1427 | int somaxconn; |
89bddce5 SH |
1428 | |
1429 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1430 | if (sock) { | |
8efa6e93 | 1431 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1432 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1433 | backlog = somaxconn; |
1da177e4 LT |
1434 | |
1435 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1436 | if (!err) |
1437 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1438 | |
6cb153ca | 1439 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1440 | } |
1441 | return err; | |
1442 | } | |
1443 | ||
1da177e4 LT |
1444 | /* |
1445 | * For accept, we attempt to create a new socket, set up the link | |
1446 | * with the client, wake up the client, then return the new | |
1447 | * connected fd. We collect the address of the connector in kernel | |
1448 | * space and move it to user at the very end. This is unclean because | |
1449 | * we open the socket then return an error. | |
1450 | * | |
1451 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1452 | * status to recvmsg. We need to add that support in a way thats | |
1453 | * clean when we restucture accept also. | |
1454 | */ | |
1455 | ||
20f37034 HC |
1456 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1457 | int __user *, upeer_addrlen, int, flags) | |
1da177e4 LT |
1458 | { |
1459 | struct socket *sock, *newsock; | |
39d8c1b6 | 1460 | struct file *newfile; |
6cb153ca | 1461 | int err, len, newfd, fput_needed; |
230b1839 | 1462 | struct sockaddr_storage address; |
1da177e4 | 1463 | |
77d27200 | 1464 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1465 | return -EINVAL; |
1466 | ||
1467 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1468 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1469 | ||
6cb153ca | 1470 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1471 | if (!sock) |
1472 | goto out; | |
1473 | ||
1474 | err = -ENFILE; | |
c6d409cf ED |
1475 | newsock = sock_alloc(); |
1476 | if (!newsock) | |
1da177e4 LT |
1477 | goto out_put; |
1478 | ||
1479 | newsock->type = sock->type; | |
1480 | newsock->ops = sock->ops; | |
1481 | ||
1da177e4 LT |
1482 | /* |
1483 | * We don't need try_module_get here, as the listening socket (sock) | |
1484 | * has the protocol module (sock->ops->owner) held. | |
1485 | */ | |
1486 | __module_get(newsock->ops->owner); | |
1487 | ||
28407630 | 1488 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1489 | if (unlikely(newfd < 0)) { |
1490 | err = newfd; | |
9a1875e6 DM |
1491 | sock_release(newsock); |
1492 | goto out_put; | |
39d8c1b6 | 1493 | } |
aab174f0 | 1494 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1495 | if (IS_ERR(newfile)) { |
28407630 AV |
1496 | err = PTR_ERR(newfile); |
1497 | put_unused_fd(newfd); | |
1498 | sock_release(newsock); | |
1499 | goto out_put; | |
1500 | } | |
39d8c1b6 | 1501 | |
a79af59e FF |
1502 | err = security_socket_accept(sock, newsock); |
1503 | if (err) | |
39d8c1b6 | 1504 | goto out_fd; |
a79af59e | 1505 | |
1da177e4 LT |
1506 | err = sock->ops->accept(sock, newsock, sock->file->f_flags); |
1507 | if (err < 0) | |
39d8c1b6 | 1508 | goto out_fd; |
1da177e4 LT |
1509 | |
1510 | if (upeer_sockaddr) { | |
230b1839 | 1511 | if (newsock->ops->getname(newsock, (struct sockaddr *)&address, |
89bddce5 | 1512 | &len, 2) < 0) { |
1da177e4 | 1513 | err = -ECONNABORTED; |
39d8c1b6 | 1514 | goto out_fd; |
1da177e4 | 1515 | } |
43db362d | 1516 | err = move_addr_to_user(&address, |
230b1839 | 1517 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1518 | if (err < 0) |
39d8c1b6 | 1519 | goto out_fd; |
1da177e4 LT |
1520 | } |
1521 | ||
1522 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1523 | ||
39d8c1b6 DM |
1524 | fd_install(newfd, newfile); |
1525 | err = newfd; | |
1da177e4 | 1526 | |
1da177e4 | 1527 | out_put: |
6cb153ca | 1528 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1529 | out: |
1530 | return err; | |
39d8c1b6 | 1531 | out_fd: |
9606a216 | 1532 | fput(newfile); |
39d8c1b6 | 1533 | put_unused_fd(newfd); |
1da177e4 LT |
1534 | goto out_put; |
1535 | } | |
1536 | ||
20f37034 HC |
1537 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1538 | int __user *, upeer_addrlen) | |
aaca0bdc | 1539 | { |
de11defe | 1540 | return sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1541 | } |
1542 | ||
1da177e4 LT |
1543 | /* |
1544 | * Attempt to connect to a socket with the server address. The address | |
1545 | * is in user space so we verify it is OK and move it to kernel space. | |
1546 | * | |
1547 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1548 | * break bindings | |
1549 | * | |
1550 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1551 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1552 | * include the -EINPROGRESS status for such sockets. | |
1553 | */ | |
1554 | ||
20f37034 HC |
1555 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1556 | int, addrlen) | |
1da177e4 LT |
1557 | { |
1558 | struct socket *sock; | |
230b1839 | 1559 | struct sockaddr_storage address; |
6cb153ca | 1560 | int err, fput_needed; |
1da177e4 | 1561 | |
6cb153ca | 1562 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1563 | if (!sock) |
1564 | goto out; | |
43db362d | 1565 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1566 | if (err < 0) |
1567 | goto out_put; | |
1568 | ||
89bddce5 | 1569 | err = |
230b1839 | 1570 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1571 | if (err) |
1572 | goto out_put; | |
1573 | ||
230b1839 | 1574 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1575 | sock->file->f_flags); |
1576 | out_put: | |
6cb153ca | 1577 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1578 | out: |
1579 | return err; | |
1580 | } | |
1581 | ||
1582 | /* | |
1583 | * Get the local address ('name') of a socket object. Move the obtained | |
1584 | * name to user space. | |
1585 | */ | |
1586 | ||
20f37034 HC |
1587 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1588 | int __user *, usockaddr_len) | |
1da177e4 LT |
1589 | { |
1590 | struct socket *sock; | |
230b1839 | 1591 | struct sockaddr_storage address; |
6cb153ca | 1592 | int len, err, fput_needed; |
89bddce5 | 1593 | |
6cb153ca | 1594 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1595 | if (!sock) |
1596 | goto out; | |
1597 | ||
1598 | err = security_socket_getsockname(sock); | |
1599 | if (err) | |
1600 | goto out_put; | |
1601 | ||
230b1839 | 1602 | err = sock->ops->getname(sock, (struct sockaddr *)&address, &len, 0); |
1da177e4 LT |
1603 | if (err) |
1604 | goto out_put; | |
43db362d | 1605 | err = move_addr_to_user(&address, len, usockaddr, usockaddr_len); |
1da177e4 LT |
1606 | |
1607 | out_put: | |
6cb153ca | 1608 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1609 | out: |
1610 | return err; | |
1611 | } | |
1612 | ||
1613 | /* | |
1614 | * Get the remote address ('name') of a socket object. Move the obtained | |
1615 | * name to user space. | |
1616 | */ | |
1617 | ||
20f37034 HC |
1618 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1619 | int __user *, usockaddr_len) | |
1da177e4 LT |
1620 | { |
1621 | struct socket *sock; | |
230b1839 | 1622 | struct sockaddr_storage address; |
6cb153ca | 1623 | int len, err, fput_needed; |
1da177e4 | 1624 | |
89bddce5 SH |
1625 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1626 | if (sock != NULL) { | |
1da177e4 LT |
1627 | err = security_socket_getpeername(sock); |
1628 | if (err) { | |
6cb153ca | 1629 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1630 | return err; |
1631 | } | |
1632 | ||
89bddce5 | 1633 | err = |
230b1839 | 1634 | sock->ops->getname(sock, (struct sockaddr *)&address, &len, |
89bddce5 | 1635 | 1); |
1da177e4 | 1636 | if (!err) |
43db362d | 1637 | err = move_addr_to_user(&address, len, usockaddr, |
89bddce5 | 1638 | usockaddr_len); |
6cb153ca | 1639 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1640 | } |
1641 | return err; | |
1642 | } | |
1643 | ||
1644 | /* | |
1645 | * Send a datagram to a given address. We move the address into kernel | |
1646 | * space and check the user space data area is readable before invoking | |
1647 | * the protocol. | |
1648 | */ | |
1649 | ||
3e0fa65f | 1650 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1651 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1652 | int, addr_len) |
1da177e4 LT |
1653 | { |
1654 | struct socket *sock; | |
230b1839 | 1655 | struct sockaddr_storage address; |
1da177e4 LT |
1656 | int err; |
1657 | struct msghdr msg; | |
1658 | struct iovec iov; | |
6cb153ca | 1659 | int fput_needed; |
6cb153ca | 1660 | |
602bd0e9 AV |
1661 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1662 | if (unlikely(err)) | |
1663 | return err; | |
de0fa95c PE |
1664 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1665 | if (!sock) | |
4387ff75 | 1666 | goto out; |
6cb153ca | 1667 | |
89bddce5 | 1668 | msg.msg_name = NULL; |
89bddce5 SH |
1669 | msg.msg_control = NULL; |
1670 | msg.msg_controllen = 0; | |
1671 | msg.msg_namelen = 0; | |
6cb153ca | 1672 | if (addr) { |
43db362d | 1673 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1674 | if (err < 0) |
1675 | goto out_put; | |
230b1839 | 1676 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1677 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1678 | } |
1679 | if (sock->file->f_flags & O_NONBLOCK) | |
1680 | flags |= MSG_DONTWAIT; | |
1681 | msg.msg_flags = flags; | |
d8725c86 | 1682 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 1683 | |
89bddce5 | 1684 | out_put: |
de0fa95c | 1685 | fput_light(sock->file, fput_needed); |
4387ff75 | 1686 | out: |
1da177e4 LT |
1687 | return err; |
1688 | } | |
1689 | ||
1690 | /* | |
89bddce5 | 1691 | * Send a datagram down a socket. |
1da177e4 LT |
1692 | */ |
1693 | ||
3e0fa65f | 1694 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1695 | unsigned int, flags) |
1da177e4 LT |
1696 | { |
1697 | return sys_sendto(fd, buff, len, flags, NULL, 0); | |
1698 | } | |
1699 | ||
1700 | /* | |
89bddce5 | 1701 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1702 | * sender. We verify the buffers are writable and if needed move the |
1703 | * sender address from kernel to user space. | |
1704 | */ | |
1705 | ||
3e0fa65f | 1706 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
95c96174 | 1707 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1708 | int __user *, addr_len) |
1da177e4 LT |
1709 | { |
1710 | struct socket *sock; | |
1711 | struct iovec iov; | |
1712 | struct msghdr msg; | |
230b1839 | 1713 | struct sockaddr_storage address; |
89bddce5 | 1714 | int err, err2; |
6cb153ca BL |
1715 | int fput_needed; |
1716 | ||
602bd0e9 AV |
1717 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
1718 | if (unlikely(err)) | |
1719 | return err; | |
de0fa95c | 1720 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 1721 | if (!sock) |
de0fa95c | 1722 | goto out; |
1da177e4 | 1723 | |
89bddce5 SH |
1724 | msg.msg_control = NULL; |
1725 | msg.msg_controllen = 0; | |
f3d33426 HFS |
1726 | /* Save some cycles and don't copy the address if not needed */ |
1727 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
1728 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
1729 | msg.msg_namelen = 0; | |
130ed5d1 | 1730 | msg.msg_iocb = NULL; |
1da177e4 LT |
1731 | if (sock->file->f_flags & O_NONBLOCK) |
1732 | flags |= MSG_DONTWAIT; | |
2da62906 | 1733 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 1734 | |
89bddce5 | 1735 | if (err >= 0 && addr != NULL) { |
43db362d | 1736 | err2 = move_addr_to_user(&address, |
230b1839 | 1737 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
1738 | if (err2 < 0) |
1739 | err = err2; | |
1da177e4 | 1740 | } |
de0fa95c PE |
1741 | |
1742 | fput_light(sock->file, fput_needed); | |
4387ff75 | 1743 | out: |
1da177e4 LT |
1744 | return err; |
1745 | } | |
1746 | ||
1747 | /* | |
89bddce5 | 1748 | * Receive a datagram from a socket. |
1da177e4 LT |
1749 | */ |
1750 | ||
b7c0ddf5 JG |
1751 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
1752 | unsigned int, flags) | |
1da177e4 LT |
1753 | { |
1754 | return sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); | |
1755 | } | |
1756 | ||
1757 | /* | |
1758 | * Set a socket option. Because we don't know the option lengths we have | |
1759 | * to pass the user mode parameter for the protocols to sort out. | |
1760 | */ | |
1761 | ||
20f37034 HC |
1762 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
1763 | char __user *, optval, int, optlen) | |
1da177e4 | 1764 | { |
6cb153ca | 1765 | int err, fput_needed; |
1da177e4 LT |
1766 | struct socket *sock; |
1767 | ||
1768 | if (optlen < 0) | |
1769 | return -EINVAL; | |
89bddce5 SH |
1770 | |
1771 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1772 | if (sock != NULL) { | |
1773 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
1774 | if (err) |
1775 | goto out_put; | |
1da177e4 LT |
1776 | |
1777 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1778 | err = |
1779 | sock_setsockopt(sock, level, optname, optval, | |
1780 | optlen); | |
1da177e4 | 1781 | else |
89bddce5 SH |
1782 | err = |
1783 | sock->ops->setsockopt(sock, level, optname, optval, | |
1784 | optlen); | |
6cb153ca BL |
1785 | out_put: |
1786 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1787 | } |
1788 | return err; | |
1789 | } | |
1790 | ||
1791 | /* | |
1792 | * Get a socket option. Because we don't know the option lengths we have | |
1793 | * to pass a user mode parameter for the protocols to sort out. | |
1794 | */ | |
1795 | ||
20f37034 HC |
1796 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
1797 | char __user *, optval, int __user *, optlen) | |
1da177e4 | 1798 | { |
6cb153ca | 1799 | int err, fput_needed; |
1da177e4 LT |
1800 | struct socket *sock; |
1801 | ||
89bddce5 SH |
1802 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1803 | if (sock != NULL) { | |
6cb153ca BL |
1804 | err = security_socket_getsockopt(sock, level, optname); |
1805 | if (err) | |
1806 | goto out_put; | |
1da177e4 LT |
1807 | |
1808 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1809 | err = |
1810 | sock_getsockopt(sock, level, optname, optval, | |
1811 | optlen); | |
1da177e4 | 1812 | else |
89bddce5 SH |
1813 | err = |
1814 | sock->ops->getsockopt(sock, level, optname, optval, | |
1815 | optlen); | |
6cb153ca BL |
1816 | out_put: |
1817 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1818 | } |
1819 | return err; | |
1820 | } | |
1821 | ||
1da177e4 LT |
1822 | /* |
1823 | * Shutdown a socket. | |
1824 | */ | |
1825 | ||
754fe8d2 | 1826 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
1da177e4 | 1827 | { |
6cb153ca | 1828 | int err, fput_needed; |
1da177e4 LT |
1829 | struct socket *sock; |
1830 | ||
89bddce5 SH |
1831 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1832 | if (sock != NULL) { | |
1da177e4 | 1833 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
1834 | if (!err) |
1835 | err = sock->ops->shutdown(sock, how); | |
1836 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1837 | } |
1838 | return err; | |
1839 | } | |
1840 | ||
89bddce5 | 1841 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
1842 | * fields which are the same type (int / unsigned) on our platforms. |
1843 | */ | |
1844 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
1845 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
1846 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
1847 | ||
c71d8ebe TH |
1848 | struct used_address { |
1849 | struct sockaddr_storage name; | |
1850 | unsigned int name_len; | |
1851 | }; | |
1852 | ||
da184284 AV |
1853 | static int copy_msghdr_from_user(struct msghdr *kmsg, |
1854 | struct user_msghdr __user *umsg, | |
1855 | struct sockaddr __user **save_addr, | |
1856 | struct iovec **iov) | |
1661bf36 | 1857 | { |
08adb7da AV |
1858 | struct sockaddr __user *uaddr; |
1859 | struct iovec __user *uiov; | |
c0371da6 | 1860 | size_t nr_segs; |
08adb7da AV |
1861 | ssize_t err; |
1862 | ||
1863 | if (!access_ok(VERIFY_READ, umsg, sizeof(*umsg)) || | |
1864 | __get_user(uaddr, &umsg->msg_name) || | |
1865 | __get_user(kmsg->msg_namelen, &umsg->msg_namelen) || | |
1866 | __get_user(uiov, &umsg->msg_iov) || | |
c0371da6 | 1867 | __get_user(nr_segs, &umsg->msg_iovlen) || |
08adb7da AV |
1868 | __get_user(kmsg->msg_control, &umsg->msg_control) || |
1869 | __get_user(kmsg->msg_controllen, &umsg->msg_controllen) || | |
1870 | __get_user(kmsg->msg_flags, &umsg->msg_flags)) | |
1661bf36 | 1871 | return -EFAULT; |
dbb490b9 | 1872 | |
08adb7da | 1873 | if (!uaddr) |
6a2a2b3a AS |
1874 | kmsg->msg_namelen = 0; |
1875 | ||
dbb490b9 ML |
1876 | if (kmsg->msg_namelen < 0) |
1877 | return -EINVAL; | |
1878 | ||
1661bf36 | 1879 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 1880 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
1881 | |
1882 | if (save_addr) | |
1883 | *save_addr = uaddr; | |
1884 | ||
1885 | if (uaddr && kmsg->msg_namelen) { | |
1886 | if (!save_addr) { | |
1887 | err = move_addr_to_kernel(uaddr, kmsg->msg_namelen, | |
1888 | kmsg->msg_name); | |
1889 | if (err < 0) | |
1890 | return err; | |
1891 | } | |
1892 | } else { | |
1893 | kmsg->msg_name = NULL; | |
1894 | kmsg->msg_namelen = 0; | |
1895 | } | |
1896 | ||
c0371da6 | 1897 | if (nr_segs > UIO_MAXIOV) |
08adb7da AV |
1898 | return -EMSGSIZE; |
1899 | ||
0345f931 | 1900 | kmsg->msg_iocb = NULL; |
1901 | ||
da184284 AV |
1902 | return import_iovec(save_addr ? READ : WRITE, uiov, nr_segs, |
1903 | UIO_FASTIOV, iov, &kmsg->msg_iter); | |
1661bf36 DC |
1904 | } |
1905 | ||
666547ff | 1906 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 1907 | struct msghdr *msg_sys, unsigned int flags, |
28a94d8f TH |
1908 | struct used_address *used_address, |
1909 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 1910 | { |
89bddce5 SH |
1911 | struct compat_msghdr __user *msg_compat = |
1912 | (struct compat_msghdr __user *)msg; | |
230b1839 | 1913 | struct sockaddr_storage address; |
1da177e4 | 1914 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 1915 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
89bddce5 SH |
1916 | __attribute__ ((aligned(sizeof(__kernel_size_t)))); |
1917 | /* 20 is size of ipv6_pktinfo */ | |
1da177e4 | 1918 | unsigned char *ctl_buf = ctl; |
d8725c86 | 1919 | int ctl_len; |
08adb7da | 1920 | ssize_t err; |
89bddce5 | 1921 | |
08adb7da | 1922 | msg_sys->msg_name = &address; |
1da177e4 | 1923 | |
08449320 | 1924 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 1925 | err = get_compat_msghdr(msg_sys, msg_compat, NULL, &iov); |
08449320 | 1926 | else |
08adb7da | 1927 | err = copy_msghdr_from_user(msg_sys, msg, NULL, &iov); |
89bddce5 | 1928 | if (err < 0) |
da184284 | 1929 | return err; |
1da177e4 LT |
1930 | |
1931 | err = -ENOBUFS; | |
1932 | ||
228e548e | 1933 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 1934 | goto out_freeiov; |
28a94d8f | 1935 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 1936 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 1937 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 1938 | err = |
228e548e | 1939 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 1940 | sizeof(ctl)); |
1da177e4 LT |
1941 | if (err) |
1942 | goto out_freeiov; | |
228e548e AB |
1943 | ctl_buf = msg_sys->msg_control; |
1944 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 1945 | } else if (ctl_len) { |
89bddce5 | 1946 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 1947 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 1948 | if (ctl_buf == NULL) |
1da177e4 LT |
1949 | goto out_freeiov; |
1950 | } | |
1951 | err = -EFAULT; | |
1952 | /* | |
228e548e | 1953 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
1954 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
1955 | * checking falls down on this. | |
1956 | */ | |
fb8621bb | 1957 | if (copy_from_user(ctl_buf, |
228e548e | 1958 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 1959 | ctl_len)) |
1da177e4 | 1960 | goto out_freectl; |
228e548e | 1961 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 1962 | } |
228e548e | 1963 | msg_sys->msg_flags = flags; |
1da177e4 LT |
1964 | |
1965 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 1966 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
1967 | /* |
1968 | * If this is sendmmsg() and current destination address is same as | |
1969 | * previously succeeded address, omit asking LSM's decision. | |
1970 | * used_address->name_len is initialized to UINT_MAX so that the first | |
1971 | * destination address never matches. | |
1972 | */ | |
bc909d9d MD |
1973 | if (used_address && msg_sys->msg_name && |
1974 | used_address->name_len == msg_sys->msg_namelen && | |
1975 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 1976 | used_address->name_len)) { |
d8725c86 | 1977 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
1978 | goto out_freectl; |
1979 | } | |
d8725c86 | 1980 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
1981 | /* |
1982 | * If this is sendmmsg() and sending to current destination address was | |
1983 | * successful, remember it. | |
1984 | */ | |
1985 | if (used_address && err >= 0) { | |
1986 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
1987 | if (msg_sys->msg_name) |
1988 | memcpy(&used_address->name, msg_sys->msg_name, | |
1989 | used_address->name_len); | |
c71d8ebe | 1990 | } |
1da177e4 LT |
1991 | |
1992 | out_freectl: | |
89bddce5 | 1993 | if (ctl_buf != ctl) |
1da177e4 LT |
1994 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
1995 | out_freeiov: | |
da184284 | 1996 | kfree(iov); |
228e548e AB |
1997 | return err; |
1998 | } | |
1999 | ||
2000 | /* | |
2001 | * BSD sendmsg interface | |
2002 | */ | |
2003 | ||
666547ff | 2004 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned flags) |
228e548e AB |
2005 | { |
2006 | int fput_needed, err; | |
2007 | struct msghdr msg_sys; | |
1be374a0 AL |
2008 | struct socket *sock; |
2009 | ||
1be374a0 | 2010 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2011 | if (!sock) |
2012 | goto out; | |
2013 | ||
28a94d8f | 2014 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2015 | |
6cb153ca | 2016 | fput_light(sock->file, fput_needed); |
89bddce5 | 2017 | out: |
1da177e4 LT |
2018 | return err; |
2019 | } | |
2020 | ||
666547ff | 2021 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 AL |
2022 | { |
2023 | if (flags & MSG_CMSG_COMPAT) | |
2024 | return -EINVAL; | |
2025 | return __sys_sendmsg(fd, msg, flags); | |
2026 | } | |
2027 | ||
228e548e AB |
2028 | /* |
2029 | * Linux sendmmsg interface | |
2030 | */ | |
2031 | ||
2032 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2033 | unsigned int flags) | |
2034 | { | |
2035 | int fput_needed, err, datagrams; | |
2036 | struct socket *sock; | |
2037 | struct mmsghdr __user *entry; | |
2038 | struct compat_mmsghdr __user *compat_entry; | |
2039 | struct msghdr msg_sys; | |
c71d8ebe | 2040 | struct used_address used_address; |
f092276d | 2041 | unsigned int oflags = flags; |
228e548e | 2042 | |
98382f41 AB |
2043 | if (vlen > UIO_MAXIOV) |
2044 | vlen = UIO_MAXIOV; | |
228e548e AB |
2045 | |
2046 | datagrams = 0; | |
2047 | ||
2048 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2049 | if (!sock) | |
2050 | return err; | |
2051 | ||
c71d8ebe | 2052 | used_address.name_len = UINT_MAX; |
228e548e AB |
2053 | entry = mmsg; |
2054 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2055 | err = 0; |
f092276d | 2056 | flags |= MSG_BATCH; |
228e548e AB |
2057 | |
2058 | while (datagrams < vlen) { | |
f092276d TH |
2059 | if (datagrams == vlen - 1) |
2060 | flags = oflags; | |
2061 | ||
228e548e | 2062 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2063 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2064 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2065 | if (err < 0) |
2066 | break; | |
2067 | err = __put_user(err, &compat_entry->msg_len); | |
2068 | ++compat_entry; | |
2069 | } else { | |
a7526eb5 | 2070 | err = ___sys_sendmsg(sock, |
666547ff | 2071 | (struct user_msghdr __user *)entry, |
28a94d8f | 2072 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2073 | if (err < 0) |
2074 | break; | |
2075 | err = put_user(err, &entry->msg_len); | |
2076 | ++entry; | |
2077 | } | |
2078 | ||
2079 | if (err) | |
2080 | break; | |
2081 | ++datagrams; | |
3023898b SHY |
2082 | if (msg_data_left(&msg_sys)) |
2083 | break; | |
a78cb84c | 2084 | cond_resched(); |
228e548e AB |
2085 | } |
2086 | ||
228e548e AB |
2087 | fput_light(sock->file, fput_needed); |
2088 | ||
728ffb86 AB |
2089 | /* We only return an error if no datagrams were able to be sent */ |
2090 | if (datagrams != 0) | |
228e548e AB |
2091 | return datagrams; |
2092 | ||
228e548e AB |
2093 | return err; |
2094 | } | |
2095 | ||
2096 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2097 | unsigned int, vlen, unsigned int, flags) | |
2098 | { | |
1be374a0 AL |
2099 | if (flags & MSG_CMSG_COMPAT) |
2100 | return -EINVAL; | |
228e548e AB |
2101 | return __sys_sendmmsg(fd, mmsg, vlen, flags); |
2102 | } | |
2103 | ||
666547ff | 2104 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2105 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2106 | { |
89bddce5 SH |
2107 | struct compat_msghdr __user *msg_compat = |
2108 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2109 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2110 | struct iovec *iov = iovstack; |
1da177e4 | 2111 | unsigned long cmsg_ptr; |
2da62906 | 2112 | int len; |
08adb7da | 2113 | ssize_t err; |
1da177e4 LT |
2114 | |
2115 | /* kernel mode address */ | |
230b1839 | 2116 | struct sockaddr_storage addr; |
1da177e4 LT |
2117 | |
2118 | /* user mode address pointers */ | |
2119 | struct sockaddr __user *uaddr; | |
08adb7da | 2120 | int __user *uaddr_len = COMPAT_NAMELEN(msg); |
89bddce5 | 2121 | |
08adb7da | 2122 | msg_sys->msg_name = &addr; |
1da177e4 | 2123 | |
f3d33426 | 2124 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2125 | err = get_compat_msghdr(msg_sys, msg_compat, &uaddr, &iov); |
f3d33426 | 2126 | else |
08adb7da | 2127 | err = copy_msghdr_from_user(msg_sys, msg, &uaddr, &iov); |
1da177e4 | 2128 | if (err < 0) |
da184284 | 2129 | return err; |
1da177e4 | 2130 | |
a2e27255 ACM |
2131 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2132 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2133 | |
f3d33426 HFS |
2134 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2135 | msg_sys->msg_namelen = 0; | |
2136 | ||
1da177e4 LT |
2137 | if (sock->file->f_flags & O_NONBLOCK) |
2138 | flags |= MSG_DONTWAIT; | |
2da62906 | 2139 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, flags); |
1da177e4 LT |
2140 | if (err < 0) |
2141 | goto out_freeiov; | |
2142 | len = err; | |
2143 | ||
2144 | if (uaddr != NULL) { | |
43db362d | 2145 | err = move_addr_to_user(&addr, |
a2e27255 | 2146 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2147 | uaddr_len); |
1da177e4 LT |
2148 | if (err < 0) |
2149 | goto out_freeiov; | |
2150 | } | |
a2e27255 | 2151 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2152 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2153 | if (err) |
2154 | goto out_freeiov; | |
2155 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2156 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2157 | &msg_compat->msg_controllen); |
2158 | else | |
a2e27255 | 2159 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2160 | &msg->msg_controllen); |
2161 | if (err) | |
2162 | goto out_freeiov; | |
2163 | err = len; | |
2164 | ||
2165 | out_freeiov: | |
da184284 | 2166 | kfree(iov); |
a2e27255 ACM |
2167 | return err; |
2168 | } | |
2169 | ||
2170 | /* | |
2171 | * BSD recvmsg interface | |
2172 | */ | |
2173 | ||
666547ff | 2174 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned flags) |
a2e27255 ACM |
2175 | { |
2176 | int fput_needed, err; | |
2177 | struct msghdr msg_sys; | |
1be374a0 AL |
2178 | struct socket *sock; |
2179 | ||
1be374a0 | 2180 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2181 | if (!sock) |
2182 | goto out; | |
2183 | ||
a7526eb5 | 2184 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2185 | |
6cb153ca | 2186 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2187 | out: |
2188 | return err; | |
2189 | } | |
2190 | ||
666547ff | 2191 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2192 | unsigned int, flags) |
2193 | { | |
2194 | if (flags & MSG_CMSG_COMPAT) | |
2195 | return -EINVAL; | |
2196 | return __sys_recvmsg(fd, msg, flags); | |
2197 | } | |
2198 | ||
a2e27255 ACM |
2199 | /* |
2200 | * Linux recvmmsg interface | |
2201 | */ | |
2202 | ||
2203 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2204 | unsigned int flags, struct timespec *timeout) | |
2205 | { | |
2206 | int fput_needed, err, datagrams; | |
2207 | struct socket *sock; | |
2208 | struct mmsghdr __user *entry; | |
d7256d0e | 2209 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2210 | struct msghdr msg_sys; |
766b9f92 DD |
2211 | struct timespec64 end_time; |
2212 | struct timespec64 timeout64; | |
a2e27255 ACM |
2213 | |
2214 | if (timeout && | |
2215 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2216 | timeout->tv_nsec)) | |
2217 | return -EINVAL; | |
2218 | ||
2219 | datagrams = 0; | |
2220 | ||
2221 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2222 | if (!sock) | |
2223 | return err; | |
2224 | ||
2225 | err = sock_error(sock->sk); | |
2226 | if (err) | |
2227 | goto out_put; | |
2228 | ||
2229 | entry = mmsg; | |
d7256d0e | 2230 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2231 | |
2232 | while (datagrams < vlen) { | |
2233 | /* | |
2234 | * No need to ask LSM for more than the first datagram. | |
2235 | */ | |
d7256d0e | 2236 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2237 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2238 | &msg_sys, flags & ~MSG_WAITFORONE, |
2239 | datagrams); | |
d7256d0e JMG |
2240 | if (err < 0) |
2241 | break; | |
2242 | err = __put_user(err, &compat_entry->msg_len); | |
2243 | ++compat_entry; | |
2244 | } else { | |
a7526eb5 | 2245 | err = ___sys_recvmsg(sock, |
666547ff | 2246 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2247 | &msg_sys, flags & ~MSG_WAITFORONE, |
2248 | datagrams); | |
d7256d0e JMG |
2249 | if (err < 0) |
2250 | break; | |
2251 | err = put_user(err, &entry->msg_len); | |
2252 | ++entry; | |
2253 | } | |
2254 | ||
a2e27255 ACM |
2255 | if (err) |
2256 | break; | |
a2e27255 ACM |
2257 | ++datagrams; |
2258 | ||
71c5c159 BB |
2259 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2260 | if (flags & MSG_WAITFORONE) | |
2261 | flags |= MSG_DONTWAIT; | |
2262 | ||
a2e27255 | 2263 | if (timeout) { |
766b9f92 DD |
2264 | ktime_get_ts64(&timeout64); |
2265 | *timeout = timespec64_to_timespec( | |
2266 | timespec64_sub(end_time, timeout64)); | |
a2e27255 ACM |
2267 | if (timeout->tv_sec < 0) { |
2268 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2269 | break; | |
2270 | } | |
2271 | ||
2272 | /* Timeout, return less than vlen datagrams */ | |
2273 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2274 | break; | |
2275 | } | |
2276 | ||
2277 | /* Out of band data, return right away */ | |
2278 | if (msg_sys.msg_flags & MSG_OOB) | |
2279 | break; | |
a78cb84c | 2280 | cond_resched(); |
a2e27255 ACM |
2281 | } |
2282 | ||
a2e27255 | 2283 | if (err == 0) |
34b88a68 ACM |
2284 | goto out_put; |
2285 | ||
2286 | if (datagrams == 0) { | |
2287 | datagrams = err; | |
2288 | goto out_put; | |
2289 | } | |
a2e27255 | 2290 | |
34b88a68 ACM |
2291 | /* |
2292 | * We may return less entries than requested (vlen) if the | |
2293 | * sock is non block and there aren't enough datagrams... | |
2294 | */ | |
2295 | if (err != -EAGAIN) { | |
a2e27255 | 2296 | /* |
34b88a68 ACM |
2297 | * ... or if recvmsg returns an error after we |
2298 | * received some datagrams, where we record the | |
2299 | * error to return on the next call or if the | |
2300 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2301 | */ |
34b88a68 | 2302 | sock->sk->sk_err = -err; |
a2e27255 | 2303 | } |
34b88a68 ACM |
2304 | out_put: |
2305 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2306 | |
34b88a68 | 2307 | return datagrams; |
a2e27255 ACM |
2308 | } |
2309 | ||
2310 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2311 | unsigned int, vlen, unsigned int, flags, | |
2312 | struct timespec __user *, timeout) | |
2313 | { | |
2314 | int datagrams; | |
2315 | struct timespec timeout_sys; | |
2316 | ||
1be374a0 AL |
2317 | if (flags & MSG_CMSG_COMPAT) |
2318 | return -EINVAL; | |
2319 | ||
a2e27255 ACM |
2320 | if (!timeout) |
2321 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2322 | ||
2323 | if (copy_from_user(&timeout_sys, timeout, sizeof(timeout_sys))) | |
2324 | return -EFAULT; | |
2325 | ||
2326 | datagrams = __sys_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
2327 | ||
2328 | if (datagrams > 0 && | |
2329 | copy_to_user(timeout, &timeout_sys, sizeof(timeout_sys))) | |
2330 | datagrams = -EFAULT; | |
2331 | ||
2332 | return datagrams; | |
2333 | } | |
2334 | ||
2335 | #ifdef __ARCH_WANT_SYS_SOCKETCALL | |
1da177e4 LT |
2336 | /* Argument list sizes for sys_socketcall */ |
2337 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2338 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2339 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2340 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2341 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2342 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2343 | }; |
2344 | ||
1da177e4 LT |
2345 | #undef AL |
2346 | ||
2347 | /* | |
89bddce5 | 2348 | * System call vectors. |
1da177e4 LT |
2349 | * |
2350 | * Argument checking cleaned up. Saved 20% in size. | |
2351 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2352 | * it is set by the callees. |
1da177e4 LT |
2353 | */ |
2354 | ||
3e0fa65f | 2355 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2356 | { |
2950fa9d | 2357 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2358 | unsigned long a0, a1; |
1da177e4 | 2359 | int err; |
47379052 | 2360 | unsigned int len; |
1da177e4 | 2361 | |
228e548e | 2362 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 LT |
2363 | return -EINVAL; |
2364 | ||
47379052 AV |
2365 | len = nargs[call]; |
2366 | if (len > sizeof(a)) | |
2367 | return -EINVAL; | |
2368 | ||
1da177e4 | 2369 | /* copy_from_user should be SMP safe. */ |
47379052 | 2370 | if (copy_from_user(a, args, len)) |
1da177e4 | 2371 | return -EFAULT; |
3ec3b2fb | 2372 | |
2950fa9d CG |
2373 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2374 | if (err) | |
2375 | return err; | |
3ec3b2fb | 2376 | |
89bddce5 SH |
2377 | a0 = a[0]; |
2378 | a1 = a[1]; | |
2379 | ||
2380 | switch (call) { | |
2381 | case SYS_SOCKET: | |
2382 | err = sys_socket(a0, a1, a[2]); | |
2383 | break; | |
2384 | case SYS_BIND: | |
2385 | err = sys_bind(a0, (struct sockaddr __user *)a1, a[2]); | |
2386 | break; | |
2387 | case SYS_CONNECT: | |
2388 | err = sys_connect(a0, (struct sockaddr __user *)a1, a[2]); | |
2389 | break; | |
2390 | case SYS_LISTEN: | |
2391 | err = sys_listen(a0, a1); | |
2392 | break; | |
2393 | case SYS_ACCEPT: | |
de11defe UD |
2394 | err = sys_accept4(a0, (struct sockaddr __user *)a1, |
2395 | (int __user *)a[2], 0); | |
89bddce5 SH |
2396 | break; |
2397 | case SYS_GETSOCKNAME: | |
2398 | err = | |
2399 | sys_getsockname(a0, (struct sockaddr __user *)a1, | |
2400 | (int __user *)a[2]); | |
2401 | break; | |
2402 | case SYS_GETPEERNAME: | |
2403 | err = | |
2404 | sys_getpeername(a0, (struct sockaddr __user *)a1, | |
2405 | (int __user *)a[2]); | |
2406 | break; | |
2407 | case SYS_SOCKETPAIR: | |
2408 | err = sys_socketpair(a0, a1, a[2], (int __user *)a[3]); | |
2409 | break; | |
2410 | case SYS_SEND: | |
2411 | err = sys_send(a0, (void __user *)a1, a[2], a[3]); | |
2412 | break; | |
2413 | case SYS_SENDTO: | |
2414 | err = sys_sendto(a0, (void __user *)a1, a[2], a[3], | |
2415 | (struct sockaddr __user *)a[4], a[5]); | |
2416 | break; | |
2417 | case SYS_RECV: | |
2418 | err = sys_recv(a0, (void __user *)a1, a[2], a[3]); | |
2419 | break; | |
2420 | case SYS_RECVFROM: | |
2421 | err = sys_recvfrom(a0, (void __user *)a1, a[2], a[3], | |
2422 | (struct sockaddr __user *)a[4], | |
2423 | (int __user *)a[5]); | |
2424 | break; | |
2425 | case SYS_SHUTDOWN: | |
2426 | err = sys_shutdown(a0, a1); | |
2427 | break; | |
2428 | case SYS_SETSOCKOPT: | |
2429 | err = sys_setsockopt(a0, a1, a[2], (char __user *)a[3], a[4]); | |
2430 | break; | |
2431 | case SYS_GETSOCKOPT: | |
2432 | err = | |
2433 | sys_getsockopt(a0, a1, a[2], (char __user *)a[3], | |
2434 | (int __user *)a[4]); | |
2435 | break; | |
2436 | case SYS_SENDMSG: | |
666547ff | 2437 | err = sys_sendmsg(a0, (struct user_msghdr __user *)a1, a[2]); |
89bddce5 | 2438 | break; |
228e548e AB |
2439 | case SYS_SENDMMSG: |
2440 | err = sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3]); | |
2441 | break; | |
89bddce5 | 2442 | case SYS_RECVMSG: |
666547ff | 2443 | err = sys_recvmsg(a0, (struct user_msghdr __user *)a1, a[2]); |
89bddce5 | 2444 | break; |
a2e27255 ACM |
2445 | case SYS_RECVMMSG: |
2446 | err = sys_recvmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3], | |
2447 | (struct timespec __user *)a[4]); | |
2448 | break; | |
de11defe UD |
2449 | case SYS_ACCEPT4: |
2450 | err = sys_accept4(a0, (struct sockaddr __user *)a1, | |
2451 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2452 | break; |
89bddce5 SH |
2453 | default: |
2454 | err = -EINVAL; | |
2455 | break; | |
1da177e4 LT |
2456 | } |
2457 | return err; | |
2458 | } | |
2459 | ||
89bddce5 | 2460 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2461 | |
55737fda SH |
2462 | /** |
2463 | * sock_register - add a socket protocol handler | |
2464 | * @ops: description of protocol | |
2465 | * | |
1da177e4 LT |
2466 | * This function is called by a protocol handler that wants to |
2467 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2468 | * socket interface. The value ops->family corresponds to the |
55737fda | 2469 | * socket system call protocol family. |
1da177e4 | 2470 | */ |
f0fd27d4 | 2471 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2472 | { |
2473 | int err; | |
2474 | ||
2475 | if (ops->family >= NPROTO) { | |
3410f22e | 2476 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2477 | return -ENOBUFS; |
2478 | } | |
55737fda SH |
2479 | |
2480 | spin_lock(&net_family_lock); | |
190683a9 ED |
2481 | if (rcu_dereference_protected(net_families[ops->family], |
2482 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2483 | err = -EEXIST; |
2484 | else { | |
cf778b00 | 2485 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2486 | err = 0; |
2487 | } | |
55737fda SH |
2488 | spin_unlock(&net_family_lock); |
2489 | ||
3410f22e | 2490 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2491 | return err; |
2492 | } | |
c6d409cf | 2493 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2494 | |
55737fda SH |
2495 | /** |
2496 | * sock_unregister - remove a protocol handler | |
2497 | * @family: protocol family to remove | |
2498 | * | |
1da177e4 LT |
2499 | * This function is called by a protocol handler that wants to |
2500 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2501 | * new socket creation. |
2502 | * | |
2503 | * If protocol handler is a module, then it can use module reference | |
2504 | * counts to protect against new references. If protocol handler is not | |
2505 | * a module then it needs to provide its own protection in | |
2506 | * the ops->create routine. | |
1da177e4 | 2507 | */ |
f0fd27d4 | 2508 | void sock_unregister(int family) |
1da177e4 | 2509 | { |
f0fd27d4 | 2510 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2511 | |
55737fda | 2512 | spin_lock(&net_family_lock); |
a9b3cd7f | 2513 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2514 | spin_unlock(&net_family_lock); |
2515 | ||
2516 | synchronize_rcu(); | |
2517 | ||
3410f22e | 2518 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2519 | } |
c6d409cf | 2520 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2521 | |
77d76ea3 | 2522 | static int __init sock_init(void) |
1da177e4 | 2523 | { |
b3e19d92 | 2524 | int err; |
2ca794e5 EB |
2525 | /* |
2526 | * Initialize the network sysctl infrastructure. | |
2527 | */ | |
2528 | err = net_sysctl_init(); | |
2529 | if (err) | |
2530 | goto out; | |
b3e19d92 | 2531 | |
1da177e4 | 2532 | /* |
89bddce5 | 2533 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2534 | */ |
2535 | skb_init(); | |
1da177e4 LT |
2536 | |
2537 | /* | |
89bddce5 | 2538 | * Initialize the protocols module. |
1da177e4 LT |
2539 | */ |
2540 | ||
2541 | init_inodecache(); | |
b3e19d92 NP |
2542 | |
2543 | err = register_filesystem(&sock_fs_type); | |
2544 | if (err) | |
2545 | goto out_fs; | |
1da177e4 | 2546 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2547 | if (IS_ERR(sock_mnt)) { |
2548 | err = PTR_ERR(sock_mnt); | |
2549 | goto out_mount; | |
2550 | } | |
77d76ea3 AK |
2551 | |
2552 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2553 | */ |
2554 | ||
2555 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
2556 | err = netfilter_init(); |
2557 | if (err) | |
2558 | goto out; | |
1da177e4 | 2559 | #endif |
cbeb321a | 2560 | |
408eccce | 2561 | ptp_classifier_init(); |
c1f19b51 | 2562 | |
b3e19d92 NP |
2563 | out: |
2564 | return err; | |
2565 | ||
2566 | out_mount: | |
2567 | unregister_filesystem(&sock_fs_type); | |
2568 | out_fs: | |
2569 | goto out; | |
1da177e4 LT |
2570 | } |
2571 | ||
77d76ea3 AK |
2572 | core_initcall(sock_init); /* early initcall */ |
2573 | ||
1da177e4 LT |
2574 | #ifdef CONFIG_PROC_FS |
2575 | void socket_seq_show(struct seq_file *seq) | |
2576 | { | |
2577 | int cpu; | |
2578 | int counter = 0; | |
2579 | ||
6f912042 | 2580 | for_each_possible_cpu(cpu) |
89bddce5 | 2581 | counter += per_cpu(sockets_in_use, cpu); |
1da177e4 LT |
2582 | |
2583 | /* It can be negative, by the way. 8) */ | |
2584 | if (counter < 0) | |
2585 | counter = 0; | |
2586 | ||
2587 | seq_printf(seq, "sockets: used %d\n", counter); | |
2588 | } | |
89bddce5 | 2589 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 2590 | |
89bbfc95 | 2591 | #ifdef CONFIG_COMPAT |
6b96018b | 2592 | static int do_siocgstamp(struct net *net, struct socket *sock, |
644595f8 | 2593 | unsigned int cmd, void __user *up) |
7a229387 | 2594 | { |
7a229387 AB |
2595 | mm_segment_t old_fs = get_fs(); |
2596 | struct timeval ktv; | |
2597 | int err; | |
2598 | ||
2599 | set_fs(KERNEL_DS); | |
6b96018b | 2600 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&ktv); |
7a229387 | 2601 | set_fs(old_fs); |
644595f8 | 2602 | if (!err) |
ed6fe9d6 | 2603 | err = compat_put_timeval(&ktv, up); |
644595f8 | 2604 | |
7a229387 AB |
2605 | return err; |
2606 | } | |
2607 | ||
6b96018b | 2608 | static int do_siocgstampns(struct net *net, struct socket *sock, |
644595f8 | 2609 | unsigned int cmd, void __user *up) |
7a229387 | 2610 | { |
7a229387 AB |
2611 | mm_segment_t old_fs = get_fs(); |
2612 | struct timespec kts; | |
2613 | int err; | |
2614 | ||
2615 | set_fs(KERNEL_DS); | |
6b96018b | 2616 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&kts); |
7a229387 | 2617 | set_fs(old_fs); |
644595f8 | 2618 | if (!err) |
ed6fe9d6 | 2619 | err = compat_put_timespec(&kts, up); |
644595f8 | 2620 | |
7a229387 AB |
2621 | return err; |
2622 | } | |
2623 | ||
6b96018b | 2624 | static int dev_ifname32(struct net *net, struct compat_ifreq __user *uifr32) |
7a229387 AB |
2625 | { |
2626 | struct ifreq __user *uifr; | |
2627 | int err; | |
2628 | ||
2629 | uifr = compat_alloc_user_space(sizeof(struct ifreq)); | |
6b96018b | 2630 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2631 | return -EFAULT; |
2632 | ||
6b96018b | 2633 | err = dev_ioctl(net, SIOCGIFNAME, uifr); |
7a229387 AB |
2634 | if (err) |
2635 | return err; | |
2636 | ||
6b96018b | 2637 | if (copy_in_user(uifr32, uifr, sizeof(struct compat_ifreq))) |
7a229387 AB |
2638 | return -EFAULT; |
2639 | ||
2640 | return 0; | |
2641 | } | |
2642 | ||
6b96018b | 2643 | static int dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 2644 | { |
6b96018b | 2645 | struct compat_ifconf ifc32; |
7a229387 AB |
2646 | struct ifconf ifc; |
2647 | struct ifconf __user *uifc; | |
6b96018b | 2648 | struct compat_ifreq __user *ifr32; |
7a229387 AB |
2649 | struct ifreq __user *ifr; |
2650 | unsigned int i, j; | |
2651 | int err; | |
2652 | ||
6b96018b | 2653 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2654 | return -EFAULT; |
2655 | ||
43da5f2e | 2656 | memset(&ifc, 0, sizeof(ifc)); |
7a229387 AB |
2657 | if (ifc32.ifcbuf == 0) { |
2658 | ifc32.ifc_len = 0; | |
2659 | ifc.ifc_len = 0; | |
2660 | ifc.ifc_req = NULL; | |
2661 | uifc = compat_alloc_user_space(sizeof(struct ifconf)); | |
2662 | } else { | |
c6d409cf ED |
2663 | size_t len = ((ifc32.ifc_len / sizeof(struct compat_ifreq)) + 1) * |
2664 | sizeof(struct ifreq); | |
7a229387 AB |
2665 | uifc = compat_alloc_user_space(sizeof(struct ifconf) + len); |
2666 | ifc.ifc_len = len; | |
2667 | ifr = ifc.ifc_req = (void __user *)(uifc + 1); | |
2668 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
c6d409cf | 2669 | for (i = 0; i < ifc32.ifc_len; i += sizeof(struct compat_ifreq)) { |
6b96018b | 2670 | if (copy_in_user(ifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2671 | return -EFAULT; |
2672 | ifr++; | |
2673 | ifr32++; | |
2674 | } | |
2675 | } | |
2676 | if (copy_to_user(uifc, &ifc, sizeof(struct ifconf))) | |
2677 | return -EFAULT; | |
2678 | ||
6b96018b | 2679 | err = dev_ioctl(net, SIOCGIFCONF, uifc); |
7a229387 AB |
2680 | if (err) |
2681 | return err; | |
2682 | ||
2683 | if (copy_from_user(&ifc, uifc, sizeof(struct ifconf))) | |
2684 | return -EFAULT; | |
2685 | ||
2686 | ifr = ifc.ifc_req; | |
2687 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
2688 | for (i = 0, j = 0; | |
c6d409cf ED |
2689 | i + sizeof(struct compat_ifreq) <= ifc32.ifc_len && j < ifc.ifc_len; |
2690 | i += sizeof(struct compat_ifreq), j += sizeof(struct ifreq)) { | |
2691 | if (copy_in_user(ifr32, ifr, sizeof(struct compat_ifreq))) | |
7a229387 AB |
2692 | return -EFAULT; |
2693 | ifr32++; | |
2694 | ifr++; | |
2695 | } | |
2696 | ||
2697 | if (ifc32.ifcbuf == 0) { | |
2698 | /* Translate from 64-bit structure multiple to | |
2699 | * a 32-bit one. | |
2700 | */ | |
2701 | i = ifc.ifc_len; | |
6b96018b | 2702 | i = ((i / sizeof(struct ifreq)) * sizeof(struct compat_ifreq)); |
7a229387 AB |
2703 | ifc32.ifc_len = i; |
2704 | } else { | |
2705 | ifc32.ifc_len = i; | |
2706 | } | |
6b96018b | 2707 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2708 | return -EFAULT; |
2709 | ||
2710 | return 0; | |
2711 | } | |
2712 | ||
6b96018b | 2713 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 2714 | { |
3a7da39d BH |
2715 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
2716 | bool convert_in = false, convert_out = false; | |
2717 | size_t buf_size = ALIGN(sizeof(struct ifreq), 8); | |
2718 | struct ethtool_rxnfc __user *rxnfc; | |
7a229387 | 2719 | struct ifreq __user *ifr; |
3a7da39d BH |
2720 | u32 rule_cnt = 0, actual_rule_cnt; |
2721 | u32 ethcmd; | |
7a229387 | 2722 | u32 data; |
3a7da39d | 2723 | int ret; |
7a229387 | 2724 | |
3a7da39d BH |
2725 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
2726 | return -EFAULT; | |
7a229387 | 2727 | |
3a7da39d BH |
2728 | compat_rxnfc = compat_ptr(data); |
2729 | ||
2730 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
2731 | return -EFAULT; |
2732 | ||
3a7da39d BH |
2733 | /* Most ethtool structures are defined without padding. |
2734 | * Unfortunately struct ethtool_rxnfc is an exception. | |
2735 | */ | |
2736 | switch (ethcmd) { | |
2737 | default: | |
2738 | break; | |
2739 | case ETHTOOL_GRXCLSRLALL: | |
2740 | /* Buffer size is variable */ | |
2741 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
2742 | return -EFAULT; | |
2743 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
2744 | return -ENOMEM; | |
2745 | buf_size += rule_cnt * sizeof(u32); | |
2746 | /* fall through */ | |
2747 | case ETHTOOL_GRXRINGS: | |
2748 | case ETHTOOL_GRXCLSRLCNT: | |
2749 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 2750 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
2751 | convert_out = true; |
2752 | /* fall through */ | |
2753 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
2754 | buf_size += sizeof(struct ethtool_rxnfc); |
2755 | convert_in = true; | |
2756 | break; | |
2757 | } | |
2758 | ||
2759 | ifr = compat_alloc_user_space(buf_size); | |
954b1244 | 2760 | rxnfc = (void __user *)ifr + ALIGN(sizeof(struct ifreq), 8); |
3a7da39d BH |
2761 | |
2762 | if (copy_in_user(&ifr->ifr_name, &ifr32->ifr_name, IFNAMSIZ)) | |
7a229387 AB |
2763 | return -EFAULT; |
2764 | ||
3a7da39d BH |
2765 | if (put_user(convert_in ? rxnfc : compat_ptr(data), |
2766 | &ifr->ifr_ifru.ifru_data)) | |
7a229387 AB |
2767 | return -EFAULT; |
2768 | ||
3a7da39d | 2769 | if (convert_in) { |
127fe533 | 2770 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
2771 | * fs.ring_cookie and at the end of fs, but nowhere else. |
2772 | */ | |
127fe533 AD |
2773 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
2774 | sizeof(compat_rxnfc->fs.m_ext) != | |
2775 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
2776 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
2777 | BUILD_BUG_ON( |
2778 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
2779 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
2780 | offsetof(struct ethtool_rxnfc, fs.location) - | |
2781 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
2782 | ||
2783 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
2784 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
2785 | (void __user *)rxnfc) || | |
3a7da39d BH |
2786 | copy_in_user(&rxnfc->fs.ring_cookie, |
2787 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 SH |
2788 | (void __user *)(&rxnfc->fs.location + 1) - |
2789 | (void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2790 | copy_in_user(&rxnfc->rule_cnt, &compat_rxnfc->rule_cnt, |
2791 | sizeof(rxnfc->rule_cnt))) | |
2792 | return -EFAULT; | |
2793 | } | |
2794 | ||
2795 | ret = dev_ioctl(net, SIOCETHTOOL, ifr); | |
2796 | if (ret) | |
2797 | return ret; | |
2798 | ||
2799 | if (convert_out) { | |
2800 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
2801 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
2802 | (const void __user *)rxnfc) || | |
3a7da39d BH |
2803 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
2804 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
2805 | (const void __user *)(&rxnfc->fs.location + 1) - |
2806 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2807 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
2808 | sizeof(rxnfc->rule_cnt))) | |
2809 | return -EFAULT; | |
2810 | ||
2811 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
2812 | /* As an optimisation, we only copy the actual | |
2813 | * number of rules that the underlying | |
2814 | * function returned. Since Mallory might | |
2815 | * change the rule count in user memory, we | |
2816 | * check that it is less than the rule count | |
2817 | * originally given (as the user buffer size), | |
2818 | * which has been range-checked. | |
2819 | */ | |
2820 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
2821 | return -EFAULT; | |
2822 | if (actual_rule_cnt < rule_cnt) | |
2823 | rule_cnt = actual_rule_cnt; | |
2824 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
2825 | &rxnfc->rule_locs[0], | |
2826 | rule_cnt * sizeof(u32))) | |
2827 | return -EFAULT; | |
2828 | } | |
2829 | } | |
2830 | ||
2831 | return 0; | |
7a229387 AB |
2832 | } |
2833 | ||
7a50a240 AB |
2834 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
2835 | { | |
2836 | void __user *uptr; | |
2837 | compat_uptr_t uptr32; | |
2838 | struct ifreq __user *uifr; | |
2839 | ||
c6d409cf | 2840 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
7a50a240 AB |
2841 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
2842 | return -EFAULT; | |
2843 | ||
2844 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
2845 | return -EFAULT; | |
2846 | ||
2847 | uptr = compat_ptr(uptr32); | |
2848 | ||
2849 | if (put_user(uptr, &uifr->ifr_settings.ifs_ifsu.raw_hdlc)) | |
2850 | return -EFAULT; | |
2851 | ||
2852 | return dev_ioctl(net, SIOCWANDEV, uifr); | |
2853 | } | |
2854 | ||
6b96018b AB |
2855 | static int bond_ioctl(struct net *net, unsigned int cmd, |
2856 | struct compat_ifreq __user *ifr32) | |
7a229387 AB |
2857 | { |
2858 | struct ifreq kifr; | |
7a229387 AB |
2859 | mm_segment_t old_fs; |
2860 | int err; | |
7a229387 AB |
2861 | |
2862 | switch (cmd) { | |
2863 | case SIOCBONDENSLAVE: | |
2864 | case SIOCBONDRELEASE: | |
2865 | case SIOCBONDSETHWADDR: | |
2866 | case SIOCBONDCHANGEACTIVE: | |
6b96018b | 2867 | if (copy_from_user(&kifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2868 | return -EFAULT; |
2869 | ||
2870 | old_fs = get_fs(); | |
c6d409cf | 2871 | set_fs(KERNEL_DS); |
c3f52ae6 | 2872 | err = dev_ioctl(net, cmd, |
2873 | (struct ifreq __user __force *) &kifr); | |
c6d409cf | 2874 | set_fs(old_fs); |
7a229387 AB |
2875 | |
2876 | return err; | |
7a229387 | 2877 | default: |
07d106d0 | 2878 | return -ENOIOCTLCMD; |
ccbd6a5a | 2879 | } |
7a229387 AB |
2880 | } |
2881 | ||
590d4693 BH |
2882 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
2883 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 2884 | struct compat_ifreq __user *u_ifreq32) |
7a229387 AB |
2885 | { |
2886 | struct ifreq __user *u_ifreq64; | |
7a229387 AB |
2887 | char tmp_buf[IFNAMSIZ]; |
2888 | void __user *data64; | |
2889 | u32 data32; | |
2890 | ||
2891 | if (copy_from_user(&tmp_buf[0], &(u_ifreq32->ifr_ifrn.ifrn_name[0]), | |
2892 | IFNAMSIZ)) | |
2893 | return -EFAULT; | |
417c3522 | 2894 | if (get_user(data32, &u_ifreq32->ifr_ifru.ifru_data)) |
7a229387 AB |
2895 | return -EFAULT; |
2896 | data64 = compat_ptr(data32); | |
2897 | ||
2898 | u_ifreq64 = compat_alloc_user_space(sizeof(*u_ifreq64)); | |
2899 | ||
7a229387 AB |
2900 | if (copy_to_user(&u_ifreq64->ifr_ifrn.ifrn_name[0], &tmp_buf[0], |
2901 | IFNAMSIZ)) | |
2902 | return -EFAULT; | |
417c3522 | 2903 | if (put_user(data64, &u_ifreq64->ifr_ifru.ifru_data)) |
7a229387 AB |
2904 | return -EFAULT; |
2905 | ||
6b96018b | 2906 | return dev_ioctl(net, cmd, u_ifreq64); |
7a229387 AB |
2907 | } |
2908 | ||
6b96018b AB |
2909 | static int dev_ifsioc(struct net *net, struct socket *sock, |
2910 | unsigned int cmd, struct compat_ifreq __user *uifr32) | |
7a229387 | 2911 | { |
a2116ed2 | 2912 | struct ifreq __user *uifr; |
7a229387 AB |
2913 | int err; |
2914 | ||
a2116ed2 AB |
2915 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
2916 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
2917 | return -EFAULT; | |
2918 | ||
2919 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
2920 | ||
7a229387 AB |
2921 | if (!err) { |
2922 | switch (cmd) { | |
2923 | case SIOCGIFFLAGS: | |
2924 | case SIOCGIFMETRIC: | |
2925 | case SIOCGIFMTU: | |
2926 | case SIOCGIFMEM: | |
2927 | case SIOCGIFHWADDR: | |
2928 | case SIOCGIFINDEX: | |
2929 | case SIOCGIFADDR: | |
2930 | case SIOCGIFBRDADDR: | |
2931 | case SIOCGIFDSTADDR: | |
2932 | case SIOCGIFNETMASK: | |
fab2532b | 2933 | case SIOCGIFPFLAGS: |
7a229387 | 2934 | case SIOCGIFTXQLEN: |
fab2532b AB |
2935 | case SIOCGMIIPHY: |
2936 | case SIOCGMIIREG: | |
a2116ed2 | 2937 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
7a229387 AB |
2938 | err = -EFAULT; |
2939 | break; | |
2940 | } | |
2941 | } | |
2942 | return err; | |
2943 | } | |
2944 | ||
a2116ed2 AB |
2945 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
2946 | struct compat_ifreq __user *uifr32) | |
2947 | { | |
2948 | struct ifreq ifr; | |
2949 | struct compat_ifmap __user *uifmap32; | |
2950 | mm_segment_t old_fs; | |
2951 | int err; | |
2952 | ||
2953 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
2954 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
2955 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
2956 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
2957 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
2958 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
2959 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
2960 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
2961 | if (err) |
2962 | return -EFAULT; | |
2963 | ||
2964 | old_fs = get_fs(); | |
c6d409cf | 2965 | set_fs(KERNEL_DS); |
c3f52ae6 | 2966 | err = dev_ioctl(net, cmd, (void __user __force *)&ifr); |
c6d409cf | 2967 | set_fs(old_fs); |
a2116ed2 AB |
2968 | |
2969 | if (cmd == SIOCGIFMAP && !err) { | |
2970 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
2971 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
2972 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
2973 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
2974 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
2975 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
2976 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
2977 | if (err) |
2978 | err = -EFAULT; | |
2979 | } | |
2980 | return err; | |
2981 | } | |
2982 | ||
7a229387 | 2983 | struct rtentry32 { |
c6d409cf | 2984 | u32 rt_pad1; |
7a229387 AB |
2985 | struct sockaddr rt_dst; /* target address */ |
2986 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
2987 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
2988 | unsigned short rt_flags; |
2989 | short rt_pad2; | |
2990 | u32 rt_pad3; | |
2991 | unsigned char rt_tos; | |
2992 | unsigned char rt_class; | |
2993 | short rt_pad4; | |
2994 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 2995 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
2996 | u32 rt_mtu; /* per route MTU/Window */ |
2997 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
2998 | unsigned short rt_irtt; /* Initial RTT */ |
2999 | }; | |
3000 | ||
3001 | struct in6_rtmsg32 { | |
3002 | struct in6_addr rtmsg_dst; | |
3003 | struct in6_addr rtmsg_src; | |
3004 | struct in6_addr rtmsg_gateway; | |
3005 | u32 rtmsg_type; | |
3006 | u16 rtmsg_dst_len; | |
3007 | u16 rtmsg_src_len; | |
3008 | u32 rtmsg_metric; | |
3009 | u32 rtmsg_info; | |
3010 | u32 rtmsg_flags; | |
3011 | s32 rtmsg_ifindex; | |
3012 | }; | |
3013 | ||
6b96018b AB |
3014 | static int routing_ioctl(struct net *net, struct socket *sock, |
3015 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3016 | { |
3017 | int ret; | |
3018 | void *r = NULL; | |
3019 | struct in6_rtmsg r6; | |
3020 | struct rtentry r4; | |
3021 | char devname[16]; | |
3022 | u32 rtdev; | |
3023 | mm_segment_t old_fs = get_fs(); | |
3024 | ||
6b96018b AB |
3025 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3026 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3027 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3028 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
3029 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3030 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3031 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3032 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3033 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3034 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3035 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3036 | |
3037 | r = (void *) &r6; | |
3038 | } else { /* ipv4 */ | |
6b96018b | 3039 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3040 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3041 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
3042 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
3043 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3044 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3045 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
3046 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3047 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3048 | if (rtdev) { |
c6d409cf | 3049 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3050 | r4.rt_dev = (char __user __force *)devname; |
3051 | devname[15] = 0; | |
7a229387 AB |
3052 | } else |
3053 | r4.rt_dev = NULL; | |
3054 | ||
3055 | r = (void *) &r4; | |
3056 | } | |
3057 | ||
3058 | if (ret) { | |
3059 | ret = -EFAULT; | |
3060 | goto out; | |
3061 | } | |
3062 | ||
c6d409cf | 3063 | set_fs(KERNEL_DS); |
6b96018b | 3064 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3065 | set_fs(old_fs); |
7a229387 AB |
3066 | |
3067 | out: | |
7a229387 AB |
3068 | return ret; |
3069 | } | |
3070 | ||
3071 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3072 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3073 | * use compatible ioctls |
7a229387 | 3074 | */ |
6b96018b | 3075 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3076 | { |
6b96018b | 3077 | compat_ulong_t tmp; |
7a229387 | 3078 | |
6b96018b | 3079 | if (get_user(tmp, argp)) |
7a229387 AB |
3080 | return -EFAULT; |
3081 | if (tmp == BRCTL_GET_VERSION) | |
3082 | return BRCTL_VERSION + 1; | |
3083 | return -EINVAL; | |
3084 | } | |
3085 | ||
6b96018b AB |
3086 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3087 | unsigned int cmd, unsigned long arg) | |
3088 | { | |
3089 | void __user *argp = compat_ptr(arg); | |
3090 | struct sock *sk = sock->sk; | |
3091 | struct net *net = sock_net(sk); | |
7a229387 | 3092 | |
6b96018b | 3093 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3094 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3095 | |
3096 | switch (cmd) { | |
3097 | case SIOCSIFBR: | |
3098 | case SIOCGIFBR: | |
3099 | return old_bridge_ioctl(argp); | |
3100 | case SIOCGIFNAME: | |
3101 | return dev_ifname32(net, argp); | |
3102 | case SIOCGIFCONF: | |
3103 | return dev_ifconf(net, argp); | |
3104 | case SIOCETHTOOL: | |
3105 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3106 | case SIOCWANDEV: |
3107 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3108 | case SIOCGIFMAP: |
3109 | case SIOCSIFMAP: | |
3110 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3111 | case SIOCBONDENSLAVE: |
3112 | case SIOCBONDRELEASE: | |
3113 | case SIOCBONDSETHWADDR: | |
6b96018b AB |
3114 | case SIOCBONDCHANGEACTIVE: |
3115 | return bond_ioctl(net, cmd, argp); | |
3116 | case SIOCADDRT: | |
3117 | case SIOCDELRT: | |
3118 | return routing_ioctl(net, sock, cmd, argp); | |
3119 | case SIOCGSTAMP: | |
3120 | return do_siocgstamp(net, sock, cmd, argp); | |
3121 | case SIOCGSTAMPNS: | |
3122 | return do_siocgstampns(net, sock, cmd, argp); | |
590d4693 BH |
3123 | case SIOCBONDSLAVEINFOQUERY: |
3124 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3125 | case SIOCSHWTSTAMP: |
fd468c74 | 3126 | case SIOCGHWTSTAMP: |
590d4693 | 3127 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3128 | |
3129 | case FIOSETOWN: | |
3130 | case SIOCSPGRP: | |
3131 | case FIOGETOWN: | |
3132 | case SIOCGPGRP: | |
3133 | case SIOCBRADDBR: | |
3134 | case SIOCBRDELBR: | |
3135 | case SIOCGIFVLAN: | |
3136 | case SIOCSIFVLAN: | |
3137 | case SIOCADDDLCI: | |
3138 | case SIOCDELDLCI: | |
c62cce2c | 3139 | case SIOCGSKNS: |
6b96018b AB |
3140 | return sock_ioctl(file, cmd, arg); |
3141 | ||
3142 | case SIOCGIFFLAGS: | |
3143 | case SIOCSIFFLAGS: | |
3144 | case SIOCGIFMETRIC: | |
3145 | case SIOCSIFMETRIC: | |
3146 | case SIOCGIFMTU: | |
3147 | case SIOCSIFMTU: | |
3148 | case SIOCGIFMEM: | |
3149 | case SIOCSIFMEM: | |
3150 | case SIOCGIFHWADDR: | |
3151 | case SIOCSIFHWADDR: | |
3152 | case SIOCADDMULTI: | |
3153 | case SIOCDELMULTI: | |
3154 | case SIOCGIFINDEX: | |
6b96018b AB |
3155 | case SIOCGIFADDR: |
3156 | case SIOCSIFADDR: | |
3157 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3158 | case SIOCDIFADDR: |
6b96018b AB |
3159 | case SIOCGIFBRDADDR: |
3160 | case SIOCSIFBRDADDR: | |
3161 | case SIOCGIFDSTADDR: | |
3162 | case SIOCSIFDSTADDR: | |
3163 | case SIOCGIFNETMASK: | |
3164 | case SIOCSIFNETMASK: | |
3165 | case SIOCSIFPFLAGS: | |
3166 | case SIOCGIFPFLAGS: | |
3167 | case SIOCGIFTXQLEN: | |
3168 | case SIOCSIFTXQLEN: | |
3169 | case SIOCBRADDIF: | |
3170 | case SIOCBRDELIF: | |
9177efd3 AB |
3171 | case SIOCSIFNAME: |
3172 | case SIOCGMIIPHY: | |
3173 | case SIOCGMIIREG: | |
3174 | case SIOCSMIIREG: | |
6b96018b | 3175 | return dev_ifsioc(net, sock, cmd, argp); |
9177efd3 | 3176 | |
6b96018b AB |
3177 | case SIOCSARP: |
3178 | case SIOCGARP: | |
3179 | case SIOCDARP: | |
6b96018b | 3180 | case SIOCATMARK: |
9177efd3 AB |
3181 | return sock_do_ioctl(net, sock, cmd, arg); |
3182 | } | |
3183 | ||
6b96018b AB |
3184 | return -ENOIOCTLCMD; |
3185 | } | |
7a229387 | 3186 | |
95c96174 | 3187 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3188 | unsigned long arg) |
89bbfc95 SP |
3189 | { |
3190 | struct socket *sock = file->private_data; | |
3191 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3192 | struct sock *sk; |
3193 | struct net *net; | |
3194 | ||
3195 | sk = sock->sk; | |
3196 | net = sock_net(sk); | |
89bbfc95 SP |
3197 | |
3198 | if (sock->ops->compat_ioctl) | |
3199 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3200 | ||
87de87d5 DM |
3201 | if (ret == -ENOIOCTLCMD && |
3202 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3203 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3204 | ||
6b96018b AB |
3205 | if (ret == -ENOIOCTLCMD) |
3206 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3207 | ||
89bbfc95 SP |
3208 | return ret; |
3209 | } | |
3210 | #endif | |
3211 | ||
ac5a488e SS |
3212 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3213 | { | |
3214 | return sock->ops->bind(sock, addr, addrlen); | |
3215 | } | |
c6d409cf | 3216 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e SS |
3217 | |
3218 | int kernel_listen(struct socket *sock, int backlog) | |
3219 | { | |
3220 | return sock->ops->listen(sock, backlog); | |
3221 | } | |
c6d409cf | 3222 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e SS |
3223 | |
3224 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) | |
3225 | { | |
3226 | struct sock *sk = sock->sk; | |
3227 | int err; | |
3228 | ||
3229 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3230 | newsock); | |
3231 | if (err < 0) | |
3232 | goto done; | |
3233 | ||
3234 | err = sock->ops->accept(sock, *newsock, flags); | |
3235 | if (err < 0) { | |
3236 | sock_release(*newsock); | |
fa8705b0 | 3237 | *newsock = NULL; |
ac5a488e SS |
3238 | goto done; |
3239 | } | |
3240 | ||
3241 | (*newsock)->ops = sock->ops; | |
1b08534e | 3242 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3243 | |
3244 | done: | |
3245 | return err; | |
3246 | } | |
c6d409cf | 3247 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e SS |
3248 | |
3249 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, | |
4768fbcb | 3250 | int flags) |
ac5a488e SS |
3251 | { |
3252 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3253 | } | |
c6d409cf | 3254 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e SS |
3255 | |
3256 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr, | |
3257 | int *addrlen) | |
3258 | { | |
3259 | return sock->ops->getname(sock, addr, addrlen, 0); | |
3260 | } | |
c6d409cf | 3261 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e SS |
3262 | |
3263 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr, | |
3264 | int *addrlen) | |
3265 | { | |
3266 | return sock->ops->getname(sock, addr, addrlen, 1); | |
3267 | } | |
c6d409cf | 3268 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e SS |
3269 | |
3270 | int kernel_getsockopt(struct socket *sock, int level, int optname, | |
3271 | char *optval, int *optlen) | |
3272 | { | |
3273 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3274 | char __user *uoptval; |
3275 | int __user *uoptlen; | |
ac5a488e SS |
3276 | int err; |
3277 | ||
fb8621bb NK |
3278 | uoptval = (char __user __force *) optval; |
3279 | uoptlen = (int __user __force *) optlen; | |
3280 | ||
ac5a488e SS |
3281 | set_fs(KERNEL_DS); |
3282 | if (level == SOL_SOCKET) | |
fb8621bb | 3283 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3284 | else |
fb8621bb NK |
3285 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3286 | uoptlen); | |
ac5a488e SS |
3287 | set_fs(oldfs); |
3288 | return err; | |
3289 | } | |
c6d409cf | 3290 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e SS |
3291 | |
3292 | int kernel_setsockopt(struct socket *sock, int level, int optname, | |
b7058842 | 3293 | char *optval, unsigned int optlen) |
ac5a488e SS |
3294 | { |
3295 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3296 | char __user *uoptval; |
ac5a488e SS |
3297 | int err; |
3298 | ||
fb8621bb NK |
3299 | uoptval = (char __user __force *) optval; |
3300 | ||
ac5a488e SS |
3301 | set_fs(KERNEL_DS); |
3302 | if (level == SOL_SOCKET) | |
fb8621bb | 3303 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3304 | else |
fb8621bb | 3305 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3306 | optlen); |
3307 | set_fs(oldfs); | |
3308 | return err; | |
3309 | } | |
c6d409cf | 3310 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e SS |
3311 | |
3312 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, | |
3313 | size_t size, int flags) | |
3314 | { | |
3315 | if (sock->ops->sendpage) | |
3316 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3317 | ||
3318 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3319 | } | |
c6d409cf | 3320 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e SS |
3321 | |
3322 | int kernel_sock_ioctl(struct socket *sock, int cmd, unsigned long arg) | |
3323 | { | |
3324 | mm_segment_t oldfs = get_fs(); | |
3325 | int err; | |
3326 | ||
3327 | set_fs(KERNEL_DS); | |
3328 | err = sock->ops->ioctl(sock, cmd, arg); | |
3329 | set_fs(oldfs); | |
3330 | ||
3331 | return err; | |
3332 | } | |
c6d409cf | 3333 | EXPORT_SYMBOL(kernel_sock_ioctl); |
ac5a488e | 3334 | |
91cf45f0 TM |
3335 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3336 | { | |
3337 | return sock->ops->shutdown(sock, how); | |
3338 | } | |
91cf45f0 | 3339 | EXPORT_SYMBOL(kernel_sock_shutdown); |