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> |
45c8178c | 92 | #include <linux/nospec.h> |
1da177e4 | 93 | |
7c0f6ba6 | 94 | #include <linux/uaccess.h> |
1da177e4 LT |
95 | #include <asm/unistd.h> |
96 | ||
97 | #include <net/compat.h> | |
87de87d5 | 98 | #include <net/wext.h> |
f8451725 | 99 | #include <net/cls_cgroup.h> |
1da177e4 LT |
100 | |
101 | #include <net/sock.h> | |
102 | #include <linux/netfilter.h> | |
103 | ||
6b96018b AB |
104 | #include <linux/if_tun.h> |
105 | #include <linux/ipv6_route.h> | |
106 | #include <linux/route.h> | |
6b96018b AB |
107 | #include <linux/sockios.h> |
108 | #include <linux/atalk.h> | |
076bb0c8 | 109 | #include <net/busy_poll.h> |
f24b9be5 | 110 | #include <linux/errqueue.h> |
06021292 | 111 | |
e0d1095a | 112 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
113 | unsigned int sysctl_net_busy_read __read_mostly; |
114 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 115 | #endif |
6b96018b | 116 | |
8ae5e030 AV |
117 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
118 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 119 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
120 | |
121 | static int sock_close(struct inode *inode, struct file *file); | |
122 | static unsigned int sock_poll(struct file *file, | |
123 | struct poll_table_struct *wait); | |
89bddce5 | 124 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
125 | #ifdef CONFIG_COMPAT |
126 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 127 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 128 | #endif |
1da177e4 | 129 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
130 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
131 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 132 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 133 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 134 | unsigned int flags); |
1da177e4 | 135 | |
1da177e4 LT |
136 | /* |
137 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
138 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
139 | */ | |
140 | ||
da7071d7 | 141 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
142 | .owner = THIS_MODULE, |
143 | .llseek = no_llseek, | |
8ae5e030 AV |
144 | .read_iter = sock_read_iter, |
145 | .write_iter = sock_write_iter, | |
1da177e4 LT |
146 | .poll = sock_poll, |
147 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
148 | #ifdef CONFIG_COMPAT |
149 | .compat_ioctl = compat_sock_ioctl, | |
150 | #endif | |
1da177e4 | 151 | .mmap = sock_mmap, |
1da177e4 LT |
152 | .release = sock_close, |
153 | .fasync = sock_fasync, | |
5274f052 JA |
154 | .sendpage = sock_sendpage, |
155 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 156 | .splice_read = sock_splice_read, |
1da177e4 LT |
157 | }; |
158 | ||
159 | /* | |
160 | * The protocol list. Each protocol is registered in here. | |
161 | */ | |
162 | ||
1da177e4 | 163 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 164 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 165 | |
1da177e4 LT |
166 | /* |
167 | * Statistics counters of the socket lists | |
168 | */ | |
169 | ||
c6d409cf | 170 | static DEFINE_PER_CPU(int, sockets_in_use); |
1da177e4 LT |
171 | |
172 | /* | |
89bddce5 SH |
173 | * Support routines. |
174 | * Move socket addresses back and forth across the kernel/user | |
175 | * divide and look after the messy bits. | |
1da177e4 LT |
176 | */ |
177 | ||
1da177e4 LT |
178 | /** |
179 | * move_addr_to_kernel - copy a socket address into kernel space | |
180 | * @uaddr: Address in user space | |
181 | * @kaddr: Address in kernel space | |
182 | * @ulen: Length in user space | |
183 | * | |
184 | * The address is copied into kernel space. If the provided address is | |
185 | * too long an error code of -EINVAL is returned. If the copy gives | |
186 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
187 | */ | |
188 | ||
43db362d | 189 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 190 | { |
230b1839 | 191 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 192 | return -EINVAL; |
89bddce5 | 193 | if (ulen == 0) |
1da177e4 | 194 | return 0; |
89bddce5 | 195 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 196 | return -EFAULT; |
3ec3b2fb | 197 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
198 | } |
199 | ||
200 | /** | |
201 | * move_addr_to_user - copy an address to user space | |
202 | * @kaddr: kernel space address | |
203 | * @klen: length of address in kernel | |
204 | * @uaddr: user space address | |
205 | * @ulen: pointer to user length field | |
206 | * | |
207 | * The value pointed to by ulen on entry is the buffer length available. | |
208 | * This is overwritten with the buffer space used. -EINVAL is returned | |
209 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
210 | * is returned if either the buffer or the length field are not | |
211 | * accessible. | |
212 | * After copying the data up to the limit the user specifies, the true | |
213 | * length of the data is written over the length limit the user | |
214 | * specified. Zero is returned for a success. | |
215 | */ | |
89bddce5 | 216 | |
43db362d | 217 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 218 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
219 | { |
220 | int err; | |
221 | int len; | |
222 | ||
68c6beb3 | 223 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
224 | err = get_user(len, ulen); |
225 | if (err) | |
1da177e4 | 226 | return err; |
89bddce5 SH |
227 | if (len > klen) |
228 | len = klen; | |
68c6beb3 | 229 | if (len < 0) |
1da177e4 | 230 | return -EINVAL; |
89bddce5 | 231 | if (len) { |
d6fe3945 SG |
232 | if (audit_sockaddr(klen, kaddr)) |
233 | return -ENOMEM; | |
89bddce5 | 234 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
235 | return -EFAULT; |
236 | } | |
237 | /* | |
89bddce5 SH |
238 | * "fromlen shall refer to the value before truncation.." |
239 | * 1003.1g | |
1da177e4 LT |
240 | */ |
241 | return __put_user(klen, ulen); | |
242 | } | |
243 | ||
e18b890b | 244 | static struct kmem_cache *sock_inode_cachep __read_mostly; |
1da177e4 LT |
245 | |
246 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
247 | { | |
248 | struct socket_alloc *ei; | |
eaefd110 | 249 | struct socket_wq *wq; |
89bddce5 | 250 | |
e94b1766 | 251 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
252 | if (!ei) |
253 | return NULL; | |
eaefd110 ED |
254 | wq = kmalloc(sizeof(*wq), GFP_KERNEL); |
255 | if (!wq) { | |
43815482 ED |
256 | kmem_cache_free(sock_inode_cachep, ei); |
257 | return NULL; | |
258 | } | |
eaefd110 ED |
259 | init_waitqueue_head(&wq->wait); |
260 | wq->fasync_list = NULL; | |
574aab1e | 261 | wq->flags = 0; |
eaefd110 | 262 | RCU_INIT_POINTER(ei->socket.wq, wq); |
89bddce5 | 263 | |
1da177e4 LT |
264 | ei->socket.state = SS_UNCONNECTED; |
265 | ei->socket.flags = 0; | |
266 | ei->socket.ops = NULL; | |
267 | ei->socket.sk = NULL; | |
268 | ei->socket.file = NULL; | |
1da177e4 LT |
269 | |
270 | return &ei->vfs_inode; | |
271 | } | |
272 | ||
273 | static void sock_destroy_inode(struct inode *inode) | |
274 | { | |
43815482 | 275 | struct socket_alloc *ei; |
eaefd110 | 276 | struct socket_wq *wq; |
43815482 ED |
277 | |
278 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
eaefd110 | 279 | wq = rcu_dereference_protected(ei->socket.wq, 1); |
61845220 | 280 | kfree_rcu(wq, rcu); |
43815482 | 281 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
282 | } |
283 | ||
51cc5068 | 284 | static void init_once(void *foo) |
1da177e4 | 285 | { |
89bddce5 | 286 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 287 | |
a35afb83 | 288 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 289 | } |
89bddce5 | 290 | |
1e911632 | 291 | static void init_inodecache(void) |
1da177e4 LT |
292 | { |
293 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
294 | sizeof(struct socket_alloc), |
295 | 0, | |
296 | (SLAB_HWCACHE_ALIGN | | |
297 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 298 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 299 | init_once); |
1e911632 | 300 | BUG_ON(sock_inode_cachep == NULL); |
1da177e4 LT |
301 | } |
302 | ||
b87221de | 303 | static const struct super_operations sockfs_ops = { |
c6d409cf ED |
304 | .alloc_inode = sock_alloc_inode, |
305 | .destroy_inode = sock_destroy_inode, | |
306 | .statfs = simple_statfs, | |
1da177e4 LT |
307 | }; |
308 | ||
c23fbb6b ED |
309 | /* |
310 | * sockfs_dname() is called from d_path(). | |
311 | */ | |
312 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
313 | { | |
314 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 315 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
316 | } |
317 | ||
3ba13d17 | 318 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 319 | .d_dname = sockfs_dname, |
1da177e4 LT |
320 | }; |
321 | ||
bba0bd31 AG |
322 | static int sockfs_xattr_get(const struct xattr_handler *handler, |
323 | struct dentry *dentry, struct inode *inode, | |
324 | const char *suffix, void *value, size_t size) | |
325 | { | |
326 | if (value) { | |
327 | if (dentry->d_name.len + 1 > size) | |
328 | return -ERANGE; | |
329 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
330 | } | |
331 | return dentry->d_name.len + 1; | |
332 | } | |
333 | ||
334 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" | |
335 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
336 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
337 | ||
338 | static const struct xattr_handler sockfs_xattr_handler = { | |
339 | .name = XATTR_NAME_SOCKPROTONAME, | |
340 | .get = sockfs_xattr_get, | |
341 | }; | |
342 | ||
4a590153 AG |
343 | static int sockfs_security_xattr_set(const struct xattr_handler *handler, |
344 | struct dentry *dentry, struct inode *inode, | |
345 | const char *suffix, const void *value, | |
346 | size_t size, int flags) | |
347 | { | |
348 | /* Handled by LSM. */ | |
349 | return -EAGAIN; | |
350 | } | |
351 | ||
352 | static const struct xattr_handler sockfs_security_xattr_handler = { | |
353 | .prefix = XATTR_SECURITY_PREFIX, | |
354 | .set = sockfs_security_xattr_set, | |
355 | }; | |
356 | ||
bba0bd31 AG |
357 | static const struct xattr_handler *sockfs_xattr_handlers[] = { |
358 | &sockfs_xattr_handler, | |
4a590153 | 359 | &sockfs_security_xattr_handler, |
bba0bd31 AG |
360 | NULL |
361 | }; | |
362 | ||
c74a1cbb AV |
363 | static struct dentry *sockfs_mount(struct file_system_type *fs_type, |
364 | int flags, const char *dev_name, void *data) | |
365 | { | |
bba0bd31 AG |
366 | return mount_pseudo_xattr(fs_type, "socket:", &sockfs_ops, |
367 | sockfs_xattr_handlers, | |
368 | &sockfs_dentry_operations, SOCKFS_MAGIC); | |
c74a1cbb AV |
369 | } |
370 | ||
371 | static struct vfsmount *sock_mnt __read_mostly; | |
372 | ||
373 | static struct file_system_type sock_fs_type = { | |
374 | .name = "sockfs", | |
375 | .mount = sockfs_mount, | |
376 | .kill_sb = kill_anon_super, | |
377 | }; | |
378 | ||
1da177e4 LT |
379 | /* |
380 | * Obtains the first available file descriptor and sets it up for use. | |
381 | * | |
39d8c1b6 DM |
382 | * These functions create file structures and maps them to fd space |
383 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
384 | * and file struct implicitly stored in sock->file. |
385 | * Note that another thread may close file descriptor before we return | |
386 | * from this function. We use the fact that now we do not refer | |
387 | * to socket after mapping. If one day we will need it, this | |
388 | * function will increment ref. count on file by 1. | |
389 | * | |
390 | * In any case returned fd MAY BE not valid! | |
391 | * This race condition is unavoidable | |
392 | * with shared fd spaces, we cannot solve it inside kernel, | |
393 | * but we take care of internal coherence yet. | |
394 | */ | |
395 | ||
aab174f0 | 396 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 397 | { |
7cbe66b6 | 398 | struct qstr name = { .name = "" }; |
2c48b9c4 | 399 | struct path path; |
7cbe66b6 | 400 | struct file *file; |
1da177e4 | 401 | |
600e1779 MY |
402 | if (dname) { |
403 | name.name = dname; | |
404 | name.len = strlen(name.name); | |
405 | } else if (sock->sk) { | |
406 | name.name = sock->sk->sk_prot_creator->name; | |
407 | name.len = strlen(name.name); | |
408 | } | |
4b936885 | 409 | path.dentry = d_alloc_pseudo(sock_mnt->mnt_sb, &name); |
28407630 AV |
410 | if (unlikely(!path.dentry)) |
411 | return ERR_PTR(-ENOMEM); | |
2c48b9c4 | 412 | path.mnt = mntget(sock_mnt); |
39d8c1b6 | 413 | |
2c48b9c4 | 414 | d_instantiate(path.dentry, SOCK_INODE(sock)); |
39d8c1b6 | 415 | |
2c48b9c4 | 416 | file = alloc_file(&path, FMODE_READ | FMODE_WRITE, |
ce8d2cdf | 417 | &socket_file_ops); |
b5ffe634 | 418 | if (IS_ERR(file)) { |
cc3808f8 | 419 | /* drop dentry, keep inode */ |
c5ef6035 | 420 | ihold(d_inode(path.dentry)); |
2c48b9c4 | 421 | path_put(&path); |
39b65252 | 422 | return file; |
cc3808f8 AV |
423 | } |
424 | ||
425 | sock->file = file; | |
77d27200 | 426 | file->f_flags = O_RDWR | (flags & O_NONBLOCK); |
39d8c1b6 | 427 | file->private_data = sock; |
28407630 | 428 | return file; |
39d8c1b6 | 429 | } |
56b31d1c | 430 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 431 | |
56b31d1c | 432 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
433 | { |
434 | struct file *newfile; | |
28407630 AV |
435 | int fd = get_unused_fd_flags(flags); |
436 | if (unlikely(fd < 0)) | |
437 | return fd; | |
39d8c1b6 | 438 | |
aab174f0 | 439 | newfile = sock_alloc_file(sock, flags, NULL); |
28407630 | 440 | if (likely(!IS_ERR(newfile))) { |
39d8c1b6 | 441 | fd_install(fd, newfile); |
28407630 AV |
442 | return fd; |
443 | } | |
7cbe66b6 | 444 | |
28407630 AV |
445 | put_unused_fd(fd); |
446 | return PTR_ERR(newfile); | |
1da177e4 LT |
447 | } |
448 | ||
406a3c63 | 449 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 450 | { |
6cb153ca BL |
451 | if (file->f_op == &socket_file_ops) |
452 | return file->private_data; /* set in sock_map_fd */ | |
453 | ||
23bb80d2 ED |
454 | *err = -ENOTSOCK; |
455 | return NULL; | |
6cb153ca | 456 | } |
406a3c63 | 457 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 458 | |
1da177e4 | 459 | /** |
c6d409cf | 460 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
461 | * @fd: file handle |
462 | * @err: pointer to an error code return | |
463 | * | |
464 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 465 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
466 | * with a negative errno code and NULL is returned. The function checks |
467 | * for both invalid handles and passing a handle which is not a socket. | |
468 | * | |
469 | * On a success the socket object pointer is returned. | |
470 | */ | |
471 | ||
472 | struct socket *sockfd_lookup(int fd, int *err) | |
473 | { | |
474 | struct file *file; | |
1da177e4 LT |
475 | struct socket *sock; |
476 | ||
89bddce5 SH |
477 | file = fget(fd); |
478 | if (!file) { | |
1da177e4 LT |
479 | *err = -EBADF; |
480 | return NULL; | |
481 | } | |
89bddce5 | 482 | |
6cb153ca BL |
483 | sock = sock_from_file(file, err); |
484 | if (!sock) | |
1da177e4 | 485 | fput(file); |
6cb153ca BL |
486 | return sock; |
487 | } | |
c6d409cf | 488 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 489 | |
6cb153ca BL |
490 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
491 | { | |
00e188ef | 492 | struct fd f = fdget(fd); |
6cb153ca BL |
493 | struct socket *sock; |
494 | ||
3672558c | 495 | *err = -EBADF; |
00e188ef AV |
496 | if (f.file) { |
497 | sock = sock_from_file(f.file, err); | |
498 | if (likely(sock)) { | |
499 | *fput_needed = f.flags; | |
6cb153ca | 500 | return sock; |
00e188ef AV |
501 | } |
502 | fdput(f); | |
1da177e4 | 503 | } |
6cb153ca | 504 | return NULL; |
1da177e4 LT |
505 | } |
506 | ||
600e1779 MY |
507 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
508 | size_t size) | |
509 | { | |
510 | ssize_t len; | |
511 | ssize_t used = 0; | |
512 | ||
c5ef6035 | 513 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
514 | if (len < 0) |
515 | return len; | |
516 | used += len; | |
517 | if (buffer) { | |
518 | if (size < used) | |
519 | return -ERANGE; | |
520 | buffer += len; | |
521 | } | |
522 | ||
523 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
524 | used += len; | |
525 | if (buffer) { | |
526 | if (size < used) | |
527 | return -ERANGE; | |
528 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
529 | buffer += len; | |
530 | } | |
531 | ||
532 | return used; | |
533 | } | |
534 | ||
dc647ec8 | 535 | static int sockfs_setattr(struct dentry *dentry, struct iattr *iattr) |
86741ec2 LC |
536 | { |
537 | int err = simple_setattr(dentry, iattr); | |
538 | ||
e1a3a60a | 539 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
540 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
541 | ||
91717ffc CW |
542 | if (sock->sk) |
543 | sock->sk->sk_uid = iattr->ia_uid; | |
544 | else | |
545 | err = -ENOENT; | |
86741ec2 LC |
546 | } |
547 | ||
548 | return err; | |
549 | } | |
550 | ||
600e1779 | 551 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 552 | .listxattr = sockfs_listxattr, |
86741ec2 | 553 | .setattr = sockfs_setattr, |
600e1779 MY |
554 | }; |
555 | ||
1da177e4 LT |
556 | /** |
557 | * sock_alloc - allocate a socket | |
89bddce5 | 558 | * |
1da177e4 LT |
559 | * Allocate a new inode and socket object. The two are bound together |
560 | * and initialised. The socket is then returned. If we are out of inodes | |
561 | * NULL is returned. | |
562 | */ | |
563 | ||
f4a00aac | 564 | struct socket *sock_alloc(void) |
1da177e4 | 565 | { |
89bddce5 SH |
566 | struct inode *inode; |
567 | struct socket *sock; | |
1da177e4 | 568 | |
a209dfc7 | 569 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
570 | if (!inode) |
571 | return NULL; | |
572 | ||
573 | sock = SOCKET_I(inode); | |
574 | ||
85fe4025 | 575 | inode->i_ino = get_next_ino(); |
89bddce5 | 576 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
577 | inode->i_uid = current_fsuid(); |
578 | inode->i_gid = current_fsgid(); | |
600e1779 | 579 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 580 | |
19e8d69c | 581 | this_cpu_add(sockets_in_use, 1); |
1da177e4 LT |
582 | return sock; |
583 | } | |
f4a00aac | 584 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 585 | |
1da177e4 LT |
586 | /** |
587 | * sock_release - close a socket | |
588 | * @sock: socket to close | |
589 | * | |
590 | * The socket is released from the protocol stack if it has a release | |
591 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 592 | * an inode not a file. |
1da177e4 | 593 | */ |
89bddce5 | 594 | |
91717ffc | 595 | static void __sock_release(struct socket *sock, struct inode *inode) |
1da177e4 LT |
596 | { |
597 | if (sock->ops) { | |
598 | struct module *owner = sock->ops->owner; | |
599 | ||
91717ffc CW |
600 | if (inode) |
601 | inode_lock(inode); | |
1da177e4 | 602 | sock->ops->release(sock); |
e5e8350d | 603 | sock->sk = NULL; |
91717ffc CW |
604 | if (inode) |
605 | inode_unlock(inode); | |
1da177e4 LT |
606 | sock->ops = NULL; |
607 | module_put(owner); | |
608 | } | |
609 | ||
eaefd110 | 610 | if (rcu_dereference_protected(sock->wq, 1)->fasync_list) |
3410f22e | 611 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 612 | |
19e8d69c | 613 | this_cpu_sub(sockets_in_use, 1); |
1da177e4 LT |
614 | if (!sock->file) { |
615 | iput(SOCK_INODE(sock)); | |
616 | return; | |
617 | } | |
89bddce5 | 618 | sock->file = NULL; |
1da177e4 | 619 | } |
91717ffc CW |
620 | |
621 | void sock_release(struct socket *sock) | |
622 | { | |
623 | __sock_release(sock, NULL); | |
624 | } | |
c6d409cf | 625 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 626 | |
c14ac945 | 627 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 628 | { |
140c55d4 ED |
629 | u8 flags = *tx_flags; |
630 | ||
c14ac945 | 631 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
632 | flags |= SKBTX_HW_TSTAMP; |
633 | ||
c14ac945 | 634 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
635 | flags |= SKBTX_SW_TSTAMP; |
636 | ||
c14ac945 | 637 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
638 | flags |= SKBTX_SCHED_TSTAMP; |
639 | ||
140c55d4 | 640 | *tx_flags = flags; |
20d49473 | 641 | } |
67cc0d40 | 642 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 643 | |
d8725c86 | 644 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 645 | { |
01e97e65 | 646 | int ret = sock->ops->sendmsg(sock, msg, msg_data_left(msg)); |
d8725c86 AV |
647 | BUG_ON(ret == -EIOCBQUEUED); |
648 | return ret; | |
1da177e4 LT |
649 | } |
650 | ||
d8725c86 | 651 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 652 | { |
d8725c86 | 653 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 654 | msg_data_left(msg)); |
228e548e | 655 | |
d8725c86 | 656 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 657 | } |
c6d409cf | 658 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 LT |
659 | |
660 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, | |
661 | struct kvec *vec, size_t num, size_t size) | |
662 | { | |
6aa24814 | 663 | iov_iter_kvec(&msg->msg_iter, WRITE | ITER_KVEC, vec, num, size); |
d8725c86 | 664 | return sock_sendmsg(sock, msg); |
1da177e4 | 665 | } |
c6d409cf | 666 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 667 | |
306b13eb TH |
668 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
669 | struct kvec *vec, size_t num, size_t size) | |
670 | { | |
671 | struct socket *sock = sk->sk_socket; | |
672 | ||
673 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 674 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb TH |
675 | |
676 | iov_iter_kvec(&msg->msg_iter, WRITE | ITER_KVEC, vec, num, size); | |
677 | ||
678 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
679 | } | |
680 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
681 | ||
8605330a SHY |
682 | static bool skb_is_err_queue(const struct sk_buff *skb) |
683 | { | |
684 | /* pkt_type of skbs enqueued on the error queue are set to | |
685 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
686 | * in recvmsg, since skbs received on a local socket will never | |
687 | * have a pkt_type of PACKET_OUTGOING. | |
688 | */ | |
689 | return skb->pkt_type == PACKET_OUTGOING; | |
690 | } | |
691 | ||
b50a5c70 ML |
692 | /* On transmit, software and hardware timestamps are returned independently. |
693 | * As the two skb clones share the hardware timestamp, which may be updated | |
694 | * before the software timestamp is received, a hardware TX timestamp may be | |
695 | * returned only if there is no software TX timestamp. Ignore false software | |
696 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
697 | * option SO_TIMESTAMP(NS) is enabled on the socket, even when the skb has a | |
698 | * hardware timestamp. | |
699 | */ | |
700 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
701 | { | |
702 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
703 | } | |
704 | ||
aad9c8c4 ML |
705 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb) |
706 | { | |
707 | struct scm_ts_pktinfo ts_pktinfo; | |
708 | struct net_device *orig_dev; | |
709 | ||
710 | if (!skb_mac_header_was_set(skb)) | |
711 | return; | |
712 | ||
713 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
714 | ||
715 | rcu_read_lock(); | |
716 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
717 | if (orig_dev) | |
718 | ts_pktinfo.if_index = orig_dev->ifindex; | |
719 | rcu_read_unlock(); | |
720 | ||
721 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
722 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
723 | sizeof(ts_pktinfo), &ts_pktinfo); | |
724 | } | |
725 | ||
92f37fd2 ED |
726 | /* |
727 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
728 | */ | |
729 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
730 | struct sk_buff *skb) | |
731 | { | |
20d49473 | 732 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
f24b9be5 | 733 | struct scm_timestamping tss; |
b50a5c70 | 734 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
735 | struct skb_shared_hwtstamps *shhwtstamps = |
736 | skb_hwtstamps(skb); | |
737 | ||
738 | /* Race occurred between timestamp enabling and packet | |
739 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 740 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 741 | __net_timestamp(skb); |
b50a5c70 ML |
742 | false_tstamp = 1; |
743 | } | |
20d49473 PO |
744 | |
745 | if (need_software_tstamp) { | |
746 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
747 | struct timeval tv; | |
748 | skb_get_timestamp(skb, &tv); | |
749 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMP, | |
750 | sizeof(tv), &tv); | |
751 | } else { | |
f24b9be5 WB |
752 | struct timespec ts; |
753 | skb_get_timestampns(skb, &ts); | |
20d49473 | 754 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPNS, |
f24b9be5 | 755 | sizeof(ts), &ts); |
20d49473 PO |
756 | } |
757 | } | |
758 | ||
f24b9be5 | 759 | memset(&tss, 0, sizeof(tss)); |
c199105d | 760 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
f24b9be5 | 761 | ktime_to_timespec_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 762 | empty = 0; |
4d276eb6 | 763 | if (shhwtstamps && |
b9f40e21 | 764 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
b50a5c70 | 765 | !skb_is_swtx_tstamp(skb, false_tstamp) && |
aad9c8c4 | 766 | ktime_to_timespec_cond(shhwtstamps->hwtstamp, tss.ts + 2)) { |
4d276eb6 | 767 | empty = 0; |
aad9c8c4 ML |
768 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && |
769 | !skb_is_err_queue(skb)) | |
770 | put_ts_pktinfo(msg, skb); | |
771 | } | |
1c885808 | 772 | if (!empty) { |
20d49473 | 773 | put_cmsg(msg, SOL_SOCKET, |
f24b9be5 | 774 | SCM_TIMESTAMPING, sizeof(tss), &tss); |
1c885808 | 775 | |
8605330a | 776 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 777 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
778 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
779 | skb->len, skb->data); | |
780 | } | |
92f37fd2 | 781 | } |
7c81fd8b ACM |
782 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
783 | ||
6e3e939f JB |
784 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
785 | struct sk_buff *skb) | |
786 | { | |
787 | int ack; | |
788 | ||
789 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
790 | return; | |
791 | if (!skb->wifi_acked_valid) | |
792 | return; | |
793 | ||
794 | ack = skb->wifi_acked; | |
795 | ||
796 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
797 | } | |
798 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
799 | ||
11165f14 | 800 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
801 | struct sk_buff *skb) | |
3b885787 | 802 | { |
744d5a3e | 803 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 804 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 805 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
806 | } |
807 | ||
767dd033 | 808 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
809 | struct sk_buff *skb) |
810 | { | |
811 | sock_recv_timestamp(msg, sk, skb); | |
812 | sock_recv_drops(msg, sk, skb); | |
813 | } | |
767dd033 | 814 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 815 | |
1b784140 | 816 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 817 | int flags) |
1da177e4 | 818 | { |
2da62906 | 819 | return sock->ops->recvmsg(sock, msg, msg_data_left(msg), flags); |
1da177e4 LT |
820 | } |
821 | ||
2da62906 | 822 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 823 | { |
2da62906 | 824 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 825 | |
2da62906 | 826 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 827 | } |
c6d409cf | 828 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 829 | |
c1249c0a ML |
830 | /** |
831 | * kernel_recvmsg - Receive a message from a socket (kernel space) | |
832 | * @sock: The socket to receive the message from | |
833 | * @msg: Received message | |
834 | * @vec: Input s/g array for message data | |
835 | * @num: Size of input s/g array | |
836 | * @size: Number of bytes to read | |
837 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
838 | * | |
839 | * On return the msg structure contains the scatter/gather array passed in the | |
840 | * vec argument. The array is modified so that it consists of the unfilled | |
841 | * portion of the original array. | |
842 | * | |
843 | * The returned value is the total number of bytes received, or an error. | |
844 | */ | |
89bddce5 SH |
845 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
846 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
847 | { |
848 | mm_segment_t oldfs = get_fs(); | |
849 | int result; | |
850 | ||
6aa24814 | 851 | iov_iter_kvec(&msg->msg_iter, READ | ITER_KVEC, vec, num, size); |
1da177e4 | 852 | set_fs(KERNEL_DS); |
2da62906 | 853 | result = sock_recvmsg(sock, msg, flags); |
1da177e4 LT |
854 | set_fs(oldfs); |
855 | return result; | |
856 | } | |
c6d409cf | 857 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 858 | |
ce1d4d3e CH |
859 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
860 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 861 | { |
1da177e4 LT |
862 | struct socket *sock; |
863 | int flags; | |
864 | ||
ce1d4d3e CH |
865 | sock = file->private_data; |
866 | ||
35f9c09f ED |
867 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
868 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
869 | flags |= more; | |
ce1d4d3e | 870 | |
e6949583 | 871 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 872 | } |
1da177e4 | 873 | |
9c55e01c | 874 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 875 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
876 | unsigned int flags) |
877 | { | |
878 | struct socket *sock = file->private_data; | |
879 | ||
997b37da RDC |
880 | if (unlikely(!sock->ops->splice_read)) |
881 | return -EINVAL; | |
882 | ||
9c55e01c JA |
883 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
884 | } | |
885 | ||
8ae5e030 | 886 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 887 | { |
6d652330 AV |
888 | struct file *file = iocb->ki_filp; |
889 | struct socket *sock = file->private_data; | |
0345f931 | 890 | struct msghdr msg = {.msg_iter = *to, |
891 | .msg_iocb = iocb}; | |
8ae5e030 | 892 | ssize_t res; |
ce1d4d3e | 893 | |
8ae5e030 AV |
894 | if (file->f_flags & O_NONBLOCK) |
895 | msg.msg_flags = MSG_DONTWAIT; | |
896 | ||
897 | if (iocb->ki_pos != 0) | |
1da177e4 | 898 | return -ESPIPE; |
027445c3 | 899 | |
66ee59af | 900 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
901 | return 0; |
902 | ||
2da62906 | 903 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
904 | *to = msg.msg_iter; |
905 | return res; | |
1da177e4 LT |
906 | } |
907 | ||
8ae5e030 | 908 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 909 | { |
6d652330 AV |
910 | struct file *file = iocb->ki_filp; |
911 | struct socket *sock = file->private_data; | |
0345f931 | 912 | struct msghdr msg = {.msg_iter = *from, |
913 | .msg_iocb = iocb}; | |
8ae5e030 | 914 | ssize_t res; |
1da177e4 | 915 | |
8ae5e030 | 916 | if (iocb->ki_pos != 0) |
ce1d4d3e | 917 | return -ESPIPE; |
027445c3 | 918 | |
8ae5e030 AV |
919 | if (file->f_flags & O_NONBLOCK) |
920 | msg.msg_flags = MSG_DONTWAIT; | |
921 | ||
6d652330 AV |
922 | if (sock->type == SOCK_SEQPACKET) |
923 | msg.msg_flags |= MSG_EOR; | |
924 | ||
d8725c86 | 925 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
926 | *from = msg.msg_iter; |
927 | return res; | |
1da177e4 LT |
928 | } |
929 | ||
1da177e4 LT |
930 | /* |
931 | * Atomic setting of ioctl hooks to avoid race | |
932 | * with module unload. | |
933 | */ | |
934 | ||
4a3e2f71 | 935 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 936 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 937 | |
881d966b | 938 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 939 | { |
4a3e2f71 | 940 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 941 | br_ioctl_hook = hook; |
4a3e2f71 | 942 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
943 | } |
944 | EXPORT_SYMBOL(brioctl_set); | |
945 | ||
4a3e2f71 | 946 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 947 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 948 | |
881d966b | 949 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 950 | { |
4a3e2f71 | 951 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 952 | vlan_ioctl_hook = hook; |
4a3e2f71 | 953 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
954 | } |
955 | EXPORT_SYMBOL(vlan_ioctl_set); | |
956 | ||
4a3e2f71 | 957 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 958 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 959 | |
89bddce5 | 960 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 961 | { |
4a3e2f71 | 962 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 963 | dlci_ioctl_hook = hook; |
4a3e2f71 | 964 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
965 | } |
966 | EXPORT_SYMBOL(dlci_ioctl_set); | |
967 | ||
6b96018b AB |
968 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
969 | unsigned int cmd, unsigned long arg) | |
970 | { | |
971 | int err; | |
972 | void __user *argp = (void __user *)arg; | |
973 | ||
974 | err = sock->ops->ioctl(sock, cmd, arg); | |
975 | ||
976 | /* | |
977 | * If this ioctl is unknown try to hand it down | |
978 | * to the NIC driver. | |
979 | */ | |
980 | if (err == -ENOIOCTLCMD) | |
981 | err = dev_ioctl(net, cmd, argp); | |
982 | ||
983 | return err; | |
984 | } | |
985 | ||
1da177e4 LT |
986 | /* |
987 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
988 | * what to do with it - that's up to the protocol still. | |
989 | */ | |
990 | ||
c62cce2c AV |
991 | static struct ns_common *get_net_ns(struct ns_common *ns) |
992 | { | |
993 | return &get_net(container_of(ns, struct net, ns))->ns; | |
994 | } | |
995 | ||
1da177e4 LT |
996 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
997 | { | |
998 | struct socket *sock; | |
881d966b | 999 | struct sock *sk; |
1da177e4 LT |
1000 | void __user *argp = (void __user *)arg; |
1001 | int pid, err; | |
881d966b | 1002 | struct net *net; |
1da177e4 | 1003 | |
b69aee04 | 1004 | sock = file->private_data; |
881d966b | 1005 | sk = sock->sk; |
3b1e0a65 | 1006 | net = sock_net(sk); |
1da177e4 | 1007 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) { |
881d966b | 1008 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 1009 | } else |
3d23e349 | 1010 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1011 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
881d966b | 1012 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 1013 | } else |
3d23e349 | 1014 | #endif |
89bddce5 | 1015 | switch (cmd) { |
1da177e4 LT |
1016 | case FIOSETOWN: |
1017 | case SIOCSPGRP: | |
1018 | err = -EFAULT; | |
1019 | if (get_user(pid, (int __user *)argp)) | |
1020 | break; | |
393cc3f5 | 1021 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1022 | break; |
1023 | case FIOGETOWN: | |
1024 | case SIOCGPGRP: | |
609d7fa9 | 1025 | err = put_user(f_getown(sock->file), |
89bddce5 | 1026 | (int __user *)argp); |
1da177e4 LT |
1027 | break; |
1028 | case SIOCGIFBR: | |
1029 | case SIOCSIFBR: | |
1030 | case SIOCBRADDBR: | |
1031 | case SIOCBRDELBR: | |
1032 | err = -ENOPKG; | |
1033 | if (!br_ioctl_hook) | |
1034 | request_module("bridge"); | |
1035 | ||
4a3e2f71 | 1036 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1037 | if (br_ioctl_hook) |
881d966b | 1038 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1039 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1040 | break; |
1041 | case SIOCGIFVLAN: | |
1042 | case SIOCSIFVLAN: | |
1043 | err = -ENOPKG; | |
1044 | if (!vlan_ioctl_hook) | |
1045 | request_module("8021q"); | |
1046 | ||
4a3e2f71 | 1047 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1048 | if (vlan_ioctl_hook) |
881d966b | 1049 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1050 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1051 | break; |
1da177e4 LT |
1052 | case SIOCADDDLCI: |
1053 | case SIOCDELDLCI: | |
1054 | err = -ENOPKG; | |
1055 | if (!dlci_ioctl_hook) | |
1056 | request_module("dlci"); | |
1057 | ||
7512cbf6 PE |
1058 | mutex_lock(&dlci_ioctl_mutex); |
1059 | if (dlci_ioctl_hook) | |
1da177e4 | 1060 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1061 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 | 1062 | break; |
c62cce2c AV |
1063 | case SIOCGSKNS: |
1064 | err = -EPERM; | |
1065 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1066 | break; | |
1067 | ||
1068 | err = open_related_ns(&net->ns, get_net_ns); | |
1069 | break; | |
1da177e4 | 1070 | default: |
6b96018b | 1071 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1072 | break; |
89bddce5 | 1073 | } |
1da177e4 LT |
1074 | return err; |
1075 | } | |
1076 | ||
1077 | int sock_create_lite(int family, int type, int protocol, struct socket **res) | |
1078 | { | |
1079 | int err; | |
1080 | struct socket *sock = NULL; | |
89bddce5 | 1081 | |
1da177e4 LT |
1082 | err = security_socket_create(family, type, protocol, 1); |
1083 | if (err) | |
1084 | goto out; | |
1085 | ||
1086 | sock = sock_alloc(); | |
1087 | if (!sock) { | |
1088 | err = -ENOMEM; | |
1089 | goto out; | |
1090 | } | |
1091 | ||
1da177e4 | 1092 | sock->type = type; |
7420ed23 VY |
1093 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1094 | if (err) | |
1095 | goto out_release; | |
1096 | ||
1da177e4 LT |
1097 | out: |
1098 | *res = sock; | |
1099 | return err; | |
7420ed23 VY |
1100 | out_release: |
1101 | sock_release(sock); | |
1102 | sock = NULL; | |
1103 | goto out; | |
1da177e4 | 1104 | } |
c6d409cf | 1105 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1106 | |
1107 | /* No kernel lock held - perfect */ | |
89bddce5 | 1108 | static unsigned int sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1109 | { |
cbf55001 | 1110 | unsigned int busy_flag = 0; |
1da177e4 LT |
1111 | struct socket *sock; |
1112 | ||
1113 | /* | |
89bddce5 | 1114 | * We can't return errors to poll, so it's either yes or no. |
1da177e4 | 1115 | */ |
b69aee04 | 1116 | sock = file->private_data; |
2d48d67f | 1117 | |
cbf55001 | 1118 | if (sk_can_busy_loop(sock->sk)) { |
2d48d67f | 1119 | /* this socket can poll_ll so tell the system call */ |
cbf55001 | 1120 | busy_flag = POLL_BUSY_LOOP; |
2d48d67f ET |
1121 | |
1122 | /* once, only if requested by syscall */ | |
cbf55001 ET |
1123 | if (wait && (wait->_key & POLL_BUSY_LOOP)) |
1124 | sk_busy_loop(sock->sk, 1); | |
2d48d67f ET |
1125 | } |
1126 | ||
cbf55001 | 1127 | return busy_flag | sock->ops->poll(file, sock, wait); |
1da177e4 LT |
1128 | } |
1129 | ||
89bddce5 | 1130 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1131 | { |
b69aee04 | 1132 | struct socket *sock = file->private_data; |
1da177e4 LT |
1133 | |
1134 | return sock->ops->mmap(file, sock, vma); | |
1135 | } | |
1136 | ||
20380731 | 1137 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1138 | { |
91717ffc | 1139 | __sock_release(SOCKET_I(inode), inode); |
1da177e4 LT |
1140 | return 0; |
1141 | } | |
1142 | ||
1143 | /* | |
1144 | * Update the socket async list | |
1145 | * | |
1146 | * Fasync_list locking strategy. | |
1147 | * | |
1148 | * 1. fasync_list is modified only under process context socket lock | |
1149 | * i.e. under semaphore. | |
1150 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1151 | * or under socket lock |
1da177e4 LT |
1152 | */ |
1153 | ||
1154 | static int sock_fasync(int fd, struct file *filp, int on) | |
1155 | { | |
989a2979 ED |
1156 | struct socket *sock = filp->private_data; |
1157 | struct sock *sk = sock->sk; | |
eaefd110 | 1158 | struct socket_wq *wq; |
1da177e4 | 1159 | |
989a2979 | 1160 | if (sk == NULL) |
1da177e4 | 1161 | return -EINVAL; |
1da177e4 LT |
1162 | |
1163 | lock_sock(sk); | |
1e1d04e6 | 1164 | wq = rcu_dereference_protected(sock->wq, lockdep_sock_is_held(sk)); |
eaefd110 | 1165 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1166 | |
eaefd110 | 1167 | if (!wq->fasync_list) |
989a2979 ED |
1168 | sock_reset_flag(sk, SOCK_FASYNC); |
1169 | else | |
bcdce719 | 1170 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1171 | |
989a2979 | 1172 | release_sock(sk); |
1da177e4 LT |
1173 | return 0; |
1174 | } | |
1175 | ||
ceb5d58b | 1176 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1177 | |
ceb5d58b | 1178 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1179 | { |
ceb5d58b | 1180 | if (!wq || !wq->fasync_list) |
1da177e4 | 1181 | return -1; |
ceb5d58b | 1182 | |
89bddce5 | 1183 | switch (how) { |
8d8ad9d7 | 1184 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1185 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1186 | break; |
1187 | goto call_kill; | |
8d8ad9d7 | 1188 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1189 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 LT |
1190 | break; |
1191 | /* fall through */ | |
8d8ad9d7 | 1192 | case SOCK_WAKE_IO: |
89bddce5 | 1193 | call_kill: |
43815482 | 1194 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1195 | break; |
8d8ad9d7 | 1196 | case SOCK_WAKE_URG: |
43815482 | 1197 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1198 | } |
ceb5d58b | 1199 | |
1da177e4 LT |
1200 | return 0; |
1201 | } | |
c6d409cf | 1202 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1203 | |
721db93a | 1204 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1205 | struct socket **res, int kern) |
1da177e4 LT |
1206 | { |
1207 | int err; | |
1208 | struct socket *sock; | |
55737fda | 1209 | const struct net_proto_family *pf; |
1da177e4 LT |
1210 | |
1211 | /* | |
89bddce5 | 1212 | * Check protocol is in range |
1da177e4 LT |
1213 | */ |
1214 | if (family < 0 || family >= NPROTO) | |
1215 | return -EAFNOSUPPORT; | |
1216 | if (type < 0 || type >= SOCK_MAX) | |
1217 | return -EINVAL; | |
1218 | ||
1219 | /* Compatibility. | |
1220 | ||
1221 | This uglymoron is moved from INET layer to here to avoid | |
1222 | deadlock in module load. | |
1223 | */ | |
1224 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1225 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1226 | current->comm); | |
1da177e4 LT |
1227 | family = PF_PACKET; |
1228 | } | |
1229 | ||
1230 | err = security_socket_create(family, type, protocol, kern); | |
1231 | if (err) | |
1232 | return err; | |
89bddce5 | 1233 | |
55737fda SH |
1234 | /* |
1235 | * Allocate the socket and allow the family to set things up. if | |
1236 | * the protocol is 0, the family is instructed to select an appropriate | |
1237 | * default. | |
1238 | */ | |
1239 | sock = sock_alloc(); | |
1240 | if (!sock) { | |
e87cc472 | 1241 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1242 | return -ENFILE; /* Not exactly a match, but its the |
1243 | closest posix thing */ | |
1244 | } | |
1245 | ||
1246 | sock->type = type; | |
1247 | ||
95a5afca | 1248 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1249 | /* Attempt to load a protocol module if the find failed. |
1250 | * | |
1251 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1252 | * requested real, full-featured networking support upon configuration. |
1253 | * Otherwise module support will break! | |
1254 | */ | |
190683a9 | 1255 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1256 | request_module("net-pf-%d", family); |
1da177e4 LT |
1257 | #endif |
1258 | ||
55737fda SH |
1259 | rcu_read_lock(); |
1260 | pf = rcu_dereference(net_families[family]); | |
1261 | err = -EAFNOSUPPORT; | |
1262 | if (!pf) | |
1263 | goto out_release; | |
1da177e4 LT |
1264 | |
1265 | /* | |
1266 | * We will call the ->create function, that possibly is in a loadable | |
1267 | * module, so we have to bump that loadable module refcnt first. | |
1268 | */ | |
55737fda | 1269 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1270 | goto out_release; |
1271 | ||
55737fda SH |
1272 | /* Now protected by module ref count */ |
1273 | rcu_read_unlock(); | |
1274 | ||
3f378b68 | 1275 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1276 | if (err < 0) |
1da177e4 | 1277 | goto out_module_put; |
a79af59e | 1278 | |
1da177e4 LT |
1279 | /* |
1280 | * Now to bump the refcnt of the [loadable] module that owns this | |
1281 | * socket at sock_release time we decrement its refcnt. | |
1282 | */ | |
55737fda SH |
1283 | if (!try_module_get(sock->ops->owner)) |
1284 | goto out_module_busy; | |
1285 | ||
1da177e4 LT |
1286 | /* |
1287 | * Now that we're done with the ->create function, the [loadable] | |
1288 | * module can have its refcnt decremented | |
1289 | */ | |
55737fda | 1290 | module_put(pf->owner); |
7420ed23 VY |
1291 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1292 | if (err) | |
3b185525 | 1293 | goto out_sock_release; |
55737fda | 1294 | *res = sock; |
1da177e4 | 1295 | |
55737fda SH |
1296 | return 0; |
1297 | ||
1298 | out_module_busy: | |
1299 | err = -EAFNOSUPPORT; | |
1da177e4 | 1300 | out_module_put: |
55737fda SH |
1301 | sock->ops = NULL; |
1302 | module_put(pf->owner); | |
1303 | out_sock_release: | |
1da177e4 | 1304 | sock_release(sock); |
55737fda SH |
1305 | return err; |
1306 | ||
1307 | out_release: | |
1308 | rcu_read_unlock(); | |
1309 | goto out_sock_release; | |
1da177e4 | 1310 | } |
721db93a | 1311 | EXPORT_SYMBOL(__sock_create); |
1da177e4 LT |
1312 | |
1313 | int sock_create(int family, int type, int protocol, struct socket **res) | |
1314 | { | |
1b8d7ae4 | 1315 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1316 | } |
c6d409cf | 1317 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1318 | |
eeb1bd5c | 1319 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1320 | { |
eeb1bd5c | 1321 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1322 | } |
c6d409cf | 1323 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1324 | |
3e0fa65f | 1325 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1da177e4 LT |
1326 | { |
1327 | int retval; | |
1328 | struct socket *sock; | |
a677a039 UD |
1329 | int flags; |
1330 | ||
e38b36f3 UD |
1331 | /* Check the SOCK_* constants for consistency. */ |
1332 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1333 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1334 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1335 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1336 | ||
a677a039 | 1337 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1338 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1339 | return -EINVAL; |
1340 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1341 | |
aaca0bdc UD |
1342 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1343 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1344 | ||
1da177e4 LT |
1345 | retval = sock_create(family, type, protocol, &sock); |
1346 | if (retval < 0) | |
1347 | goto out; | |
1348 | ||
77d27200 | 1349 | retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1350 | if (retval < 0) |
1351 | goto out_release; | |
1352 | ||
1353 | out: | |
1354 | /* It may be already another descriptor 8) Not kernel problem. */ | |
1355 | return retval; | |
1356 | ||
1357 | out_release: | |
1358 | sock_release(sock); | |
1359 | return retval; | |
1360 | } | |
1361 | ||
1362 | /* | |
1363 | * Create a pair of connected sockets. | |
1364 | */ | |
1365 | ||
3e0fa65f HC |
1366 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1367 | int __user *, usockvec) | |
1da177e4 LT |
1368 | { |
1369 | struct socket *sock1, *sock2; | |
1370 | int fd1, fd2, err; | |
db349509 | 1371 | struct file *newfile1, *newfile2; |
a677a039 UD |
1372 | int flags; |
1373 | ||
1374 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1375 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1376 | return -EINVAL; |
1377 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1378 | |
aaca0bdc UD |
1379 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1380 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1381 | ||
1da177e4 LT |
1382 | /* |
1383 | * Obtain the first socket and check if the underlying protocol | |
1384 | * supports the socketpair call. | |
1385 | */ | |
1386 | ||
1387 | err = sock_create(family, type, protocol, &sock1); | |
1388 | if (err < 0) | |
1389 | goto out; | |
1390 | ||
1391 | err = sock_create(family, type, protocol, &sock2); | |
1392 | if (err < 0) | |
1393 | goto out_release_1; | |
1394 | ||
1395 | err = sock1->ops->socketpair(sock1, sock2); | |
89bddce5 | 1396 | if (err < 0) |
1da177e4 LT |
1397 | goto out_release_both; |
1398 | ||
28407630 | 1399 | fd1 = get_unused_fd_flags(flags); |
bf3c23d1 DM |
1400 | if (unlikely(fd1 < 0)) { |
1401 | err = fd1; | |
db349509 | 1402 | goto out_release_both; |
bf3c23d1 | 1403 | } |
d73aa286 | 1404 | |
28407630 | 1405 | fd2 = get_unused_fd_flags(flags); |
198de4d7 AV |
1406 | if (unlikely(fd2 < 0)) { |
1407 | err = fd2; | |
d73aa286 | 1408 | goto out_put_unused_1; |
28407630 AV |
1409 | } |
1410 | ||
aab174f0 | 1411 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1412 | if (IS_ERR(newfile1)) { |
28407630 | 1413 | err = PTR_ERR(newfile1); |
d73aa286 | 1414 | goto out_put_unused_both; |
28407630 AV |
1415 | } |
1416 | ||
aab174f0 | 1417 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1418 | if (IS_ERR(newfile2)) { |
1419 | err = PTR_ERR(newfile2); | |
d73aa286 | 1420 | goto out_fput_1; |
db349509 AV |
1421 | } |
1422 | ||
d73aa286 YD |
1423 | err = put_user(fd1, &usockvec[0]); |
1424 | if (err) | |
1425 | goto out_fput_both; | |
1426 | ||
1427 | err = put_user(fd2, &usockvec[1]); | |
1428 | if (err) | |
1429 | goto out_fput_both; | |
1430 | ||
157cf649 | 1431 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1432 | |
db349509 AV |
1433 | fd_install(fd1, newfile1); |
1434 | fd_install(fd2, newfile2); | |
1da177e4 LT |
1435 | /* fd1 and fd2 may be already another descriptors. |
1436 | * Not kernel problem. | |
1437 | */ | |
1438 | ||
d73aa286 | 1439 | return 0; |
1da177e4 | 1440 | |
d73aa286 YD |
1441 | out_fput_both: |
1442 | fput(newfile2); | |
1443 | fput(newfile1); | |
1444 | put_unused_fd(fd2); | |
1445 | put_unused_fd(fd1); | |
1446 | goto out; | |
1447 | ||
1448 | out_fput_1: | |
1449 | fput(newfile1); | |
1450 | put_unused_fd(fd2); | |
1451 | put_unused_fd(fd1); | |
1452 | sock_release(sock2); | |
1453 | goto out; | |
1da177e4 | 1454 | |
d73aa286 YD |
1455 | out_put_unused_both: |
1456 | put_unused_fd(fd2); | |
1457 | out_put_unused_1: | |
1458 | put_unused_fd(fd1); | |
1da177e4 | 1459 | out_release_both: |
89bddce5 | 1460 | sock_release(sock2); |
1da177e4 | 1461 | out_release_1: |
89bddce5 | 1462 | sock_release(sock1); |
1da177e4 LT |
1463 | out: |
1464 | return err; | |
1465 | } | |
1466 | ||
1da177e4 LT |
1467 | /* |
1468 | * Bind a name to a socket. Nothing much to do here since it's | |
1469 | * the protocol's responsibility to handle the local address. | |
1470 | * | |
1471 | * We move the socket address to kernel space before we call | |
1472 | * the protocol layer (having also checked the address is ok). | |
1473 | */ | |
1474 | ||
20f37034 | 1475 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1da177e4 LT |
1476 | { |
1477 | struct socket *sock; | |
230b1839 | 1478 | struct sockaddr_storage address; |
6cb153ca | 1479 | int err, fput_needed; |
1da177e4 | 1480 | |
89bddce5 | 1481 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1482 | if (sock) { |
43db362d | 1483 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
89bddce5 SH |
1484 | if (err >= 0) { |
1485 | err = security_socket_bind(sock, | |
230b1839 | 1486 | (struct sockaddr *)&address, |
89bddce5 | 1487 | addrlen); |
6cb153ca BL |
1488 | if (!err) |
1489 | err = sock->ops->bind(sock, | |
89bddce5 | 1490 | (struct sockaddr *) |
230b1839 | 1491 | &address, addrlen); |
1da177e4 | 1492 | } |
6cb153ca | 1493 | fput_light(sock->file, fput_needed); |
89bddce5 | 1494 | } |
1da177e4 LT |
1495 | return err; |
1496 | } | |
1497 | ||
1da177e4 LT |
1498 | /* |
1499 | * Perform a listen. Basically, we allow the protocol to do anything | |
1500 | * necessary for a listen, and if that works, we mark the socket as | |
1501 | * ready for listening. | |
1502 | */ | |
1503 | ||
3e0fa65f | 1504 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1da177e4 LT |
1505 | { |
1506 | struct socket *sock; | |
6cb153ca | 1507 | int err, fput_needed; |
b8e1f9b5 | 1508 | int somaxconn; |
89bddce5 SH |
1509 | |
1510 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1511 | if (sock) { | |
8efa6e93 | 1512 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1513 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1514 | backlog = somaxconn; |
1da177e4 LT |
1515 | |
1516 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1517 | if (!err) |
1518 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1519 | |
6cb153ca | 1520 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1521 | } |
1522 | return err; | |
1523 | } | |
1524 | ||
1da177e4 LT |
1525 | /* |
1526 | * For accept, we attempt to create a new socket, set up the link | |
1527 | * with the client, wake up the client, then return the new | |
1528 | * connected fd. We collect the address of the connector in kernel | |
1529 | * space and move it to user at the very end. This is unclean because | |
1530 | * we open the socket then return an error. | |
1531 | * | |
1532 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1533 | * status to recvmsg. We need to add that support in a way thats | |
1534 | * clean when we restucture accept also. | |
1535 | */ | |
1536 | ||
20f37034 HC |
1537 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1538 | int __user *, upeer_addrlen, int, flags) | |
1da177e4 LT |
1539 | { |
1540 | struct socket *sock, *newsock; | |
39d8c1b6 | 1541 | struct file *newfile; |
6cb153ca | 1542 | int err, len, newfd, fput_needed; |
230b1839 | 1543 | struct sockaddr_storage address; |
1da177e4 | 1544 | |
77d27200 | 1545 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1546 | return -EINVAL; |
1547 | ||
1548 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1549 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1550 | ||
6cb153ca | 1551 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1552 | if (!sock) |
1553 | goto out; | |
1554 | ||
1555 | err = -ENFILE; | |
c6d409cf ED |
1556 | newsock = sock_alloc(); |
1557 | if (!newsock) | |
1da177e4 LT |
1558 | goto out_put; |
1559 | ||
1560 | newsock->type = sock->type; | |
1561 | newsock->ops = sock->ops; | |
1562 | ||
1da177e4 LT |
1563 | /* |
1564 | * We don't need try_module_get here, as the listening socket (sock) | |
1565 | * has the protocol module (sock->ops->owner) held. | |
1566 | */ | |
1567 | __module_get(newsock->ops->owner); | |
1568 | ||
28407630 | 1569 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1570 | if (unlikely(newfd < 0)) { |
1571 | err = newfd; | |
9a1875e6 DM |
1572 | sock_release(newsock); |
1573 | goto out_put; | |
39d8c1b6 | 1574 | } |
aab174f0 | 1575 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1576 | if (IS_ERR(newfile)) { |
28407630 AV |
1577 | err = PTR_ERR(newfile); |
1578 | put_unused_fd(newfd); | |
1579 | sock_release(newsock); | |
1580 | goto out_put; | |
1581 | } | |
39d8c1b6 | 1582 | |
a79af59e FF |
1583 | err = security_socket_accept(sock, newsock); |
1584 | if (err) | |
39d8c1b6 | 1585 | goto out_fd; |
a79af59e | 1586 | |
cdfbabfb | 1587 | err = sock->ops->accept(sock, newsock, sock->file->f_flags, false); |
1da177e4 | 1588 | if (err < 0) |
39d8c1b6 | 1589 | goto out_fd; |
1da177e4 LT |
1590 | |
1591 | if (upeer_sockaddr) { | |
230b1839 | 1592 | if (newsock->ops->getname(newsock, (struct sockaddr *)&address, |
89bddce5 | 1593 | &len, 2) < 0) { |
1da177e4 | 1594 | err = -ECONNABORTED; |
39d8c1b6 | 1595 | goto out_fd; |
1da177e4 | 1596 | } |
43db362d | 1597 | err = move_addr_to_user(&address, |
230b1839 | 1598 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1599 | if (err < 0) |
39d8c1b6 | 1600 | goto out_fd; |
1da177e4 LT |
1601 | } |
1602 | ||
1603 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1604 | ||
39d8c1b6 DM |
1605 | fd_install(newfd, newfile); |
1606 | err = newfd; | |
1da177e4 | 1607 | |
1da177e4 | 1608 | out_put: |
6cb153ca | 1609 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1610 | out: |
1611 | return err; | |
39d8c1b6 | 1612 | out_fd: |
9606a216 | 1613 | fput(newfile); |
39d8c1b6 | 1614 | put_unused_fd(newfd); |
1da177e4 LT |
1615 | goto out_put; |
1616 | } | |
1617 | ||
20f37034 HC |
1618 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1619 | int __user *, upeer_addrlen) | |
aaca0bdc | 1620 | { |
de11defe | 1621 | return sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1622 | } |
1623 | ||
1da177e4 LT |
1624 | /* |
1625 | * Attempt to connect to a socket with the server address. The address | |
1626 | * is in user space so we verify it is OK and move it to kernel space. | |
1627 | * | |
1628 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1629 | * break bindings | |
1630 | * | |
1631 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1632 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1633 | * include the -EINPROGRESS status for such sockets. | |
1634 | */ | |
1635 | ||
20f37034 HC |
1636 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1637 | int, addrlen) | |
1da177e4 LT |
1638 | { |
1639 | struct socket *sock; | |
230b1839 | 1640 | struct sockaddr_storage address; |
6cb153ca | 1641 | int err, fput_needed; |
1da177e4 | 1642 | |
6cb153ca | 1643 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1644 | if (!sock) |
1645 | goto out; | |
43db362d | 1646 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1647 | if (err < 0) |
1648 | goto out_put; | |
1649 | ||
89bddce5 | 1650 | err = |
230b1839 | 1651 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1652 | if (err) |
1653 | goto out_put; | |
1654 | ||
230b1839 | 1655 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1656 | sock->file->f_flags); |
1657 | out_put: | |
6cb153ca | 1658 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1659 | out: |
1660 | return err; | |
1661 | } | |
1662 | ||
1663 | /* | |
1664 | * Get the local address ('name') of a socket object. Move the obtained | |
1665 | * name to user space. | |
1666 | */ | |
1667 | ||
20f37034 HC |
1668 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1669 | int __user *, usockaddr_len) | |
1da177e4 LT |
1670 | { |
1671 | struct socket *sock; | |
230b1839 | 1672 | struct sockaddr_storage address; |
6cb153ca | 1673 | int len, err, fput_needed; |
89bddce5 | 1674 | |
6cb153ca | 1675 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1676 | if (!sock) |
1677 | goto out; | |
1678 | ||
1679 | err = security_socket_getsockname(sock); | |
1680 | if (err) | |
1681 | goto out_put; | |
1682 | ||
230b1839 | 1683 | err = sock->ops->getname(sock, (struct sockaddr *)&address, &len, 0); |
1da177e4 LT |
1684 | if (err) |
1685 | goto out_put; | |
43db362d | 1686 | err = move_addr_to_user(&address, len, usockaddr, usockaddr_len); |
1da177e4 LT |
1687 | |
1688 | out_put: | |
6cb153ca | 1689 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1690 | out: |
1691 | return err; | |
1692 | } | |
1693 | ||
1694 | /* | |
1695 | * Get the remote address ('name') of a socket object. Move the obtained | |
1696 | * name to user space. | |
1697 | */ | |
1698 | ||
20f37034 HC |
1699 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1700 | int __user *, usockaddr_len) | |
1da177e4 LT |
1701 | { |
1702 | struct socket *sock; | |
230b1839 | 1703 | struct sockaddr_storage address; |
6cb153ca | 1704 | int len, err, fput_needed; |
1da177e4 | 1705 | |
89bddce5 SH |
1706 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1707 | if (sock != NULL) { | |
1da177e4 LT |
1708 | err = security_socket_getpeername(sock); |
1709 | if (err) { | |
6cb153ca | 1710 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1711 | return err; |
1712 | } | |
1713 | ||
89bddce5 | 1714 | err = |
230b1839 | 1715 | sock->ops->getname(sock, (struct sockaddr *)&address, &len, |
89bddce5 | 1716 | 1); |
1da177e4 | 1717 | if (!err) |
43db362d | 1718 | err = move_addr_to_user(&address, len, usockaddr, |
89bddce5 | 1719 | usockaddr_len); |
6cb153ca | 1720 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1721 | } |
1722 | return err; | |
1723 | } | |
1724 | ||
1725 | /* | |
1726 | * Send a datagram to a given address. We move the address into kernel | |
1727 | * space and check the user space data area is readable before invoking | |
1728 | * the protocol. | |
1729 | */ | |
1730 | ||
3e0fa65f | 1731 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1732 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1733 | int, addr_len) |
1da177e4 LT |
1734 | { |
1735 | struct socket *sock; | |
230b1839 | 1736 | struct sockaddr_storage address; |
1da177e4 LT |
1737 | int err; |
1738 | struct msghdr msg; | |
1739 | struct iovec iov; | |
6cb153ca | 1740 | int fput_needed; |
6cb153ca | 1741 | |
602bd0e9 AV |
1742 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1743 | if (unlikely(err)) | |
1744 | return err; | |
de0fa95c PE |
1745 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1746 | if (!sock) | |
4387ff75 | 1747 | goto out; |
6cb153ca | 1748 | |
89bddce5 | 1749 | msg.msg_name = NULL; |
89bddce5 SH |
1750 | msg.msg_control = NULL; |
1751 | msg.msg_controllen = 0; | |
1752 | msg.msg_namelen = 0; | |
6cb153ca | 1753 | if (addr) { |
43db362d | 1754 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1755 | if (err < 0) |
1756 | goto out_put; | |
230b1839 | 1757 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1758 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1759 | } |
1760 | if (sock->file->f_flags & O_NONBLOCK) | |
1761 | flags |= MSG_DONTWAIT; | |
1762 | msg.msg_flags = flags; | |
d8725c86 | 1763 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 1764 | |
89bddce5 | 1765 | out_put: |
de0fa95c | 1766 | fput_light(sock->file, fput_needed); |
4387ff75 | 1767 | out: |
1da177e4 LT |
1768 | return err; |
1769 | } | |
1770 | ||
1771 | /* | |
89bddce5 | 1772 | * Send a datagram down a socket. |
1da177e4 LT |
1773 | */ |
1774 | ||
3e0fa65f | 1775 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1776 | unsigned int, flags) |
1da177e4 LT |
1777 | { |
1778 | return sys_sendto(fd, buff, len, flags, NULL, 0); | |
1779 | } | |
1780 | ||
1781 | /* | |
89bddce5 | 1782 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1783 | * sender. We verify the buffers are writable and if needed move the |
1784 | * sender address from kernel to user space. | |
1785 | */ | |
1786 | ||
3e0fa65f | 1787 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
95c96174 | 1788 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1789 | int __user *, addr_len) |
1da177e4 LT |
1790 | { |
1791 | struct socket *sock; | |
1792 | struct iovec iov; | |
1793 | struct msghdr msg; | |
230b1839 | 1794 | struct sockaddr_storage address; |
89bddce5 | 1795 | int err, err2; |
6cb153ca BL |
1796 | int fput_needed; |
1797 | ||
602bd0e9 AV |
1798 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
1799 | if (unlikely(err)) | |
1800 | return err; | |
de0fa95c | 1801 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 1802 | if (!sock) |
de0fa95c | 1803 | goto out; |
1da177e4 | 1804 | |
89bddce5 SH |
1805 | msg.msg_control = NULL; |
1806 | msg.msg_controllen = 0; | |
f3d33426 HFS |
1807 | /* Save some cycles and don't copy the address if not needed */ |
1808 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
1809 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
1810 | msg.msg_namelen = 0; | |
130ed5d1 | 1811 | msg.msg_iocb = NULL; |
9f138fa6 | 1812 | msg.msg_flags = 0; |
1da177e4 LT |
1813 | if (sock->file->f_flags & O_NONBLOCK) |
1814 | flags |= MSG_DONTWAIT; | |
2da62906 | 1815 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 1816 | |
89bddce5 | 1817 | if (err >= 0 && addr != NULL) { |
43db362d | 1818 | err2 = move_addr_to_user(&address, |
230b1839 | 1819 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
1820 | if (err2 < 0) |
1821 | err = err2; | |
1da177e4 | 1822 | } |
de0fa95c PE |
1823 | |
1824 | fput_light(sock->file, fput_needed); | |
4387ff75 | 1825 | out: |
1da177e4 LT |
1826 | return err; |
1827 | } | |
1828 | ||
1829 | /* | |
89bddce5 | 1830 | * Receive a datagram from a socket. |
1da177e4 LT |
1831 | */ |
1832 | ||
b7c0ddf5 JG |
1833 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
1834 | unsigned int, flags) | |
1da177e4 LT |
1835 | { |
1836 | return sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); | |
1837 | } | |
1838 | ||
1839 | /* | |
1840 | * Set a socket option. Because we don't know the option lengths we have | |
1841 | * to pass the user mode parameter for the protocols to sort out. | |
1842 | */ | |
1843 | ||
20f37034 HC |
1844 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
1845 | char __user *, optval, int, optlen) | |
1da177e4 | 1846 | { |
6cb153ca | 1847 | int err, fput_needed; |
1da177e4 LT |
1848 | struct socket *sock; |
1849 | ||
1850 | if (optlen < 0) | |
1851 | return -EINVAL; | |
89bddce5 SH |
1852 | |
1853 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1854 | if (sock != NULL) { | |
1855 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
1856 | if (err) |
1857 | goto out_put; | |
1da177e4 LT |
1858 | |
1859 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1860 | err = |
1861 | sock_setsockopt(sock, level, optname, optval, | |
1862 | optlen); | |
1da177e4 | 1863 | else |
89bddce5 SH |
1864 | err = |
1865 | sock->ops->setsockopt(sock, level, optname, optval, | |
1866 | optlen); | |
6cb153ca BL |
1867 | out_put: |
1868 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1869 | } |
1870 | return err; | |
1871 | } | |
1872 | ||
1873 | /* | |
1874 | * Get a socket option. Because we don't know the option lengths we have | |
1875 | * to pass a user mode parameter for the protocols to sort out. | |
1876 | */ | |
1877 | ||
20f37034 HC |
1878 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
1879 | char __user *, optval, int __user *, optlen) | |
1da177e4 | 1880 | { |
6cb153ca | 1881 | int err, fput_needed; |
1da177e4 LT |
1882 | struct socket *sock; |
1883 | ||
89bddce5 SH |
1884 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1885 | if (sock != NULL) { | |
6cb153ca BL |
1886 | err = security_socket_getsockopt(sock, level, optname); |
1887 | if (err) | |
1888 | goto out_put; | |
1da177e4 LT |
1889 | |
1890 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1891 | err = |
1892 | sock_getsockopt(sock, level, optname, optval, | |
1893 | optlen); | |
1da177e4 | 1894 | else |
89bddce5 SH |
1895 | err = |
1896 | sock->ops->getsockopt(sock, level, optname, optval, | |
1897 | optlen); | |
6cb153ca BL |
1898 | out_put: |
1899 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1900 | } |
1901 | return err; | |
1902 | } | |
1903 | ||
1da177e4 LT |
1904 | /* |
1905 | * Shutdown a socket. | |
1906 | */ | |
1907 | ||
754fe8d2 | 1908 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
1da177e4 | 1909 | { |
6cb153ca | 1910 | int err, fput_needed; |
1da177e4 LT |
1911 | struct socket *sock; |
1912 | ||
89bddce5 SH |
1913 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1914 | if (sock != NULL) { | |
1da177e4 | 1915 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
1916 | if (!err) |
1917 | err = sock->ops->shutdown(sock, how); | |
1918 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1919 | } |
1920 | return err; | |
1921 | } | |
1922 | ||
89bddce5 | 1923 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
1924 | * fields which are the same type (int / unsigned) on our platforms. |
1925 | */ | |
1926 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
1927 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
1928 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
1929 | ||
c71d8ebe TH |
1930 | struct used_address { |
1931 | struct sockaddr_storage name; | |
1932 | unsigned int name_len; | |
1933 | }; | |
1934 | ||
da184284 AV |
1935 | static int copy_msghdr_from_user(struct msghdr *kmsg, |
1936 | struct user_msghdr __user *umsg, | |
1937 | struct sockaddr __user **save_addr, | |
1938 | struct iovec **iov) | |
1661bf36 | 1939 | { |
ffb07550 | 1940 | struct user_msghdr msg; |
08adb7da AV |
1941 | ssize_t err; |
1942 | ||
ffb07550 | 1943 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 1944 | return -EFAULT; |
dbb490b9 | 1945 | |
864d9664 | 1946 | kmsg->msg_control = (void __force *)msg.msg_control; |
ffb07550 AV |
1947 | kmsg->msg_controllen = msg.msg_controllen; |
1948 | kmsg->msg_flags = msg.msg_flags; | |
1949 | ||
1950 | kmsg->msg_namelen = msg.msg_namelen; | |
1951 | if (!msg.msg_name) | |
6a2a2b3a AS |
1952 | kmsg->msg_namelen = 0; |
1953 | ||
dbb490b9 ML |
1954 | if (kmsg->msg_namelen < 0) |
1955 | return -EINVAL; | |
1956 | ||
1661bf36 | 1957 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 1958 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
1959 | |
1960 | if (save_addr) | |
ffb07550 | 1961 | *save_addr = msg.msg_name; |
08adb7da | 1962 | |
ffb07550 | 1963 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 1964 | if (!save_addr) { |
864d9664 PA |
1965 | err = move_addr_to_kernel(msg.msg_name, |
1966 | kmsg->msg_namelen, | |
08adb7da AV |
1967 | kmsg->msg_name); |
1968 | if (err < 0) | |
1969 | return err; | |
1970 | } | |
1971 | } else { | |
1972 | kmsg->msg_name = NULL; | |
1973 | kmsg->msg_namelen = 0; | |
1974 | } | |
1975 | ||
ffb07550 | 1976 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
1977 | return -EMSGSIZE; |
1978 | ||
0345f931 | 1979 | kmsg->msg_iocb = NULL; |
1980 | ||
ffb07550 AV |
1981 | return import_iovec(save_addr ? READ : WRITE, |
1982 | msg.msg_iov, msg.msg_iovlen, | |
da184284 | 1983 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
1661bf36 DC |
1984 | } |
1985 | ||
666547ff | 1986 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 1987 | struct msghdr *msg_sys, unsigned int flags, |
28a94d8f TH |
1988 | struct used_address *used_address, |
1989 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 1990 | { |
89bddce5 SH |
1991 | struct compat_msghdr __user *msg_compat = |
1992 | (struct compat_msghdr __user *)msg; | |
230b1839 | 1993 | struct sockaddr_storage address; |
1da177e4 | 1994 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 1995 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 1996 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 1997 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 1998 | unsigned char *ctl_buf = ctl; |
d8725c86 | 1999 | int ctl_len; |
08adb7da | 2000 | ssize_t err; |
89bddce5 | 2001 | |
08adb7da | 2002 | msg_sys->msg_name = &address; |
1da177e4 | 2003 | |
08449320 | 2004 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2005 | err = get_compat_msghdr(msg_sys, msg_compat, NULL, &iov); |
08449320 | 2006 | else |
08adb7da | 2007 | err = copy_msghdr_from_user(msg_sys, msg, NULL, &iov); |
89bddce5 | 2008 | if (err < 0) |
da184284 | 2009 | return err; |
1da177e4 LT |
2010 | |
2011 | err = -ENOBUFS; | |
2012 | ||
228e548e | 2013 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 2014 | goto out_freeiov; |
28a94d8f | 2015 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2016 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2017 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2018 | err = |
228e548e | 2019 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2020 | sizeof(ctl)); |
1da177e4 LT |
2021 | if (err) |
2022 | goto out_freeiov; | |
228e548e AB |
2023 | ctl_buf = msg_sys->msg_control; |
2024 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2025 | } else if (ctl_len) { |
ac4340fc DM |
2026 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2027 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2028 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2029 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2030 | if (ctl_buf == NULL) |
1da177e4 LT |
2031 | goto out_freeiov; |
2032 | } | |
2033 | err = -EFAULT; | |
2034 | /* | |
228e548e | 2035 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2036 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2037 | * checking falls down on this. | |
2038 | */ | |
fb8621bb | 2039 | if (copy_from_user(ctl_buf, |
228e548e | 2040 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2041 | ctl_len)) |
1da177e4 | 2042 | goto out_freectl; |
228e548e | 2043 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2044 | } |
228e548e | 2045 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2046 | |
2047 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2048 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2049 | /* |
2050 | * If this is sendmmsg() and current destination address is same as | |
2051 | * previously succeeded address, omit asking LSM's decision. | |
2052 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2053 | * destination address never matches. | |
2054 | */ | |
bc909d9d MD |
2055 | if (used_address && msg_sys->msg_name && |
2056 | used_address->name_len == msg_sys->msg_namelen && | |
2057 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2058 | used_address->name_len)) { |
d8725c86 | 2059 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2060 | goto out_freectl; |
2061 | } | |
d8725c86 | 2062 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2063 | /* |
2064 | * If this is sendmmsg() and sending to current destination address was | |
2065 | * successful, remember it. | |
2066 | */ | |
2067 | if (used_address && err >= 0) { | |
2068 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2069 | if (msg_sys->msg_name) |
2070 | memcpy(&used_address->name, msg_sys->msg_name, | |
2071 | used_address->name_len); | |
c71d8ebe | 2072 | } |
1da177e4 LT |
2073 | |
2074 | out_freectl: | |
89bddce5 | 2075 | if (ctl_buf != ctl) |
1da177e4 LT |
2076 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
2077 | out_freeiov: | |
da184284 | 2078 | kfree(iov); |
228e548e AB |
2079 | return err; |
2080 | } | |
2081 | ||
2082 | /* | |
2083 | * BSD sendmsg interface | |
2084 | */ | |
2085 | ||
666547ff | 2086 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned flags) |
228e548e AB |
2087 | { |
2088 | int fput_needed, err; | |
2089 | struct msghdr msg_sys; | |
1be374a0 AL |
2090 | struct socket *sock; |
2091 | ||
1be374a0 | 2092 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2093 | if (!sock) |
2094 | goto out; | |
2095 | ||
28a94d8f | 2096 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2097 | |
6cb153ca | 2098 | fput_light(sock->file, fput_needed); |
89bddce5 | 2099 | out: |
1da177e4 LT |
2100 | return err; |
2101 | } | |
2102 | ||
666547ff | 2103 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 AL |
2104 | { |
2105 | if (flags & MSG_CMSG_COMPAT) | |
2106 | return -EINVAL; | |
2107 | return __sys_sendmsg(fd, msg, flags); | |
2108 | } | |
2109 | ||
228e548e AB |
2110 | /* |
2111 | * Linux sendmmsg interface | |
2112 | */ | |
2113 | ||
2114 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2115 | unsigned int flags) | |
2116 | { | |
2117 | int fput_needed, err, datagrams; | |
2118 | struct socket *sock; | |
2119 | struct mmsghdr __user *entry; | |
2120 | struct compat_mmsghdr __user *compat_entry; | |
2121 | struct msghdr msg_sys; | |
c71d8ebe | 2122 | struct used_address used_address; |
f092276d | 2123 | unsigned int oflags = flags; |
228e548e | 2124 | |
98382f41 AB |
2125 | if (vlen > UIO_MAXIOV) |
2126 | vlen = UIO_MAXIOV; | |
228e548e AB |
2127 | |
2128 | datagrams = 0; | |
2129 | ||
2130 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2131 | if (!sock) | |
2132 | return err; | |
2133 | ||
c71d8ebe | 2134 | used_address.name_len = UINT_MAX; |
228e548e AB |
2135 | entry = mmsg; |
2136 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2137 | err = 0; |
f092276d | 2138 | flags |= MSG_BATCH; |
228e548e AB |
2139 | |
2140 | while (datagrams < vlen) { | |
f092276d TH |
2141 | if (datagrams == vlen - 1) |
2142 | flags = oflags; | |
2143 | ||
228e548e | 2144 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2145 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2146 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2147 | if (err < 0) |
2148 | break; | |
2149 | err = __put_user(err, &compat_entry->msg_len); | |
2150 | ++compat_entry; | |
2151 | } else { | |
a7526eb5 | 2152 | err = ___sys_sendmsg(sock, |
666547ff | 2153 | (struct user_msghdr __user *)entry, |
28a94d8f | 2154 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2155 | if (err < 0) |
2156 | break; | |
2157 | err = put_user(err, &entry->msg_len); | |
2158 | ++entry; | |
2159 | } | |
2160 | ||
2161 | if (err) | |
2162 | break; | |
2163 | ++datagrams; | |
3023898b SHY |
2164 | if (msg_data_left(&msg_sys)) |
2165 | break; | |
a78cb84c | 2166 | cond_resched(); |
228e548e AB |
2167 | } |
2168 | ||
228e548e AB |
2169 | fput_light(sock->file, fput_needed); |
2170 | ||
728ffb86 AB |
2171 | /* We only return an error if no datagrams were able to be sent */ |
2172 | if (datagrams != 0) | |
228e548e AB |
2173 | return datagrams; |
2174 | ||
228e548e AB |
2175 | return err; |
2176 | } | |
2177 | ||
2178 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2179 | unsigned int, vlen, unsigned int, flags) | |
2180 | { | |
1be374a0 AL |
2181 | if (flags & MSG_CMSG_COMPAT) |
2182 | return -EINVAL; | |
228e548e AB |
2183 | return __sys_sendmmsg(fd, mmsg, vlen, flags); |
2184 | } | |
2185 | ||
666547ff | 2186 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2187 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2188 | { |
89bddce5 SH |
2189 | struct compat_msghdr __user *msg_compat = |
2190 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2191 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2192 | struct iovec *iov = iovstack; |
1da177e4 | 2193 | unsigned long cmsg_ptr; |
2da62906 | 2194 | int len; |
08adb7da | 2195 | ssize_t err; |
1da177e4 LT |
2196 | |
2197 | /* kernel mode address */ | |
230b1839 | 2198 | struct sockaddr_storage addr; |
1da177e4 LT |
2199 | |
2200 | /* user mode address pointers */ | |
2201 | struct sockaddr __user *uaddr; | |
08adb7da | 2202 | int __user *uaddr_len = COMPAT_NAMELEN(msg); |
89bddce5 | 2203 | |
08adb7da | 2204 | msg_sys->msg_name = &addr; |
1da177e4 | 2205 | |
f3d33426 | 2206 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2207 | err = get_compat_msghdr(msg_sys, msg_compat, &uaddr, &iov); |
f3d33426 | 2208 | else |
08adb7da | 2209 | err = copy_msghdr_from_user(msg_sys, msg, &uaddr, &iov); |
1da177e4 | 2210 | if (err < 0) |
da184284 | 2211 | return err; |
1da177e4 | 2212 | |
a2e27255 ACM |
2213 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2214 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2215 | |
f3d33426 HFS |
2216 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2217 | msg_sys->msg_namelen = 0; | |
2218 | ||
1da177e4 LT |
2219 | if (sock->file->f_flags & O_NONBLOCK) |
2220 | flags |= MSG_DONTWAIT; | |
2da62906 | 2221 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, flags); |
1da177e4 LT |
2222 | if (err < 0) |
2223 | goto out_freeiov; | |
2224 | len = err; | |
2225 | ||
2226 | if (uaddr != NULL) { | |
43db362d | 2227 | err = move_addr_to_user(&addr, |
a2e27255 | 2228 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2229 | uaddr_len); |
1da177e4 LT |
2230 | if (err < 0) |
2231 | goto out_freeiov; | |
2232 | } | |
a2e27255 | 2233 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2234 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2235 | if (err) |
2236 | goto out_freeiov; | |
2237 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2238 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2239 | &msg_compat->msg_controllen); |
2240 | else | |
a2e27255 | 2241 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2242 | &msg->msg_controllen); |
2243 | if (err) | |
2244 | goto out_freeiov; | |
2245 | err = len; | |
2246 | ||
2247 | out_freeiov: | |
da184284 | 2248 | kfree(iov); |
a2e27255 ACM |
2249 | return err; |
2250 | } | |
2251 | ||
2252 | /* | |
2253 | * BSD recvmsg interface | |
2254 | */ | |
2255 | ||
666547ff | 2256 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned flags) |
a2e27255 ACM |
2257 | { |
2258 | int fput_needed, err; | |
2259 | struct msghdr msg_sys; | |
1be374a0 AL |
2260 | struct socket *sock; |
2261 | ||
1be374a0 | 2262 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2263 | if (!sock) |
2264 | goto out; | |
2265 | ||
a7526eb5 | 2266 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2267 | |
6cb153ca | 2268 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2269 | out: |
2270 | return err; | |
2271 | } | |
2272 | ||
666547ff | 2273 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2274 | unsigned int, flags) |
2275 | { | |
2276 | if (flags & MSG_CMSG_COMPAT) | |
2277 | return -EINVAL; | |
2278 | return __sys_recvmsg(fd, msg, flags); | |
2279 | } | |
2280 | ||
a2e27255 ACM |
2281 | /* |
2282 | * Linux recvmmsg interface | |
2283 | */ | |
2284 | ||
2285 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2286 | unsigned int flags, struct timespec *timeout) | |
2287 | { | |
2288 | int fput_needed, err, datagrams; | |
2289 | struct socket *sock; | |
2290 | struct mmsghdr __user *entry; | |
d7256d0e | 2291 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2292 | struct msghdr msg_sys; |
766b9f92 DD |
2293 | struct timespec64 end_time; |
2294 | struct timespec64 timeout64; | |
a2e27255 ACM |
2295 | |
2296 | if (timeout && | |
2297 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2298 | timeout->tv_nsec)) | |
2299 | return -EINVAL; | |
2300 | ||
2301 | datagrams = 0; | |
2302 | ||
2303 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2304 | if (!sock) | |
2305 | return err; | |
2306 | ||
2307 | err = sock_error(sock->sk); | |
e623a9e9 MJ |
2308 | if (err) { |
2309 | datagrams = err; | |
a2e27255 | 2310 | goto out_put; |
e623a9e9 | 2311 | } |
a2e27255 ACM |
2312 | |
2313 | entry = mmsg; | |
d7256d0e | 2314 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2315 | |
2316 | while (datagrams < vlen) { | |
2317 | /* | |
2318 | * No need to ask LSM for more than the first datagram. | |
2319 | */ | |
d7256d0e | 2320 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2321 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2322 | &msg_sys, flags & ~MSG_WAITFORONE, |
2323 | datagrams); | |
d7256d0e JMG |
2324 | if (err < 0) |
2325 | break; | |
2326 | err = __put_user(err, &compat_entry->msg_len); | |
2327 | ++compat_entry; | |
2328 | } else { | |
a7526eb5 | 2329 | err = ___sys_recvmsg(sock, |
666547ff | 2330 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2331 | &msg_sys, flags & ~MSG_WAITFORONE, |
2332 | datagrams); | |
d7256d0e JMG |
2333 | if (err < 0) |
2334 | break; | |
2335 | err = put_user(err, &entry->msg_len); | |
2336 | ++entry; | |
2337 | } | |
2338 | ||
a2e27255 ACM |
2339 | if (err) |
2340 | break; | |
a2e27255 ACM |
2341 | ++datagrams; |
2342 | ||
71c5c159 BB |
2343 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2344 | if (flags & MSG_WAITFORONE) | |
2345 | flags |= MSG_DONTWAIT; | |
2346 | ||
a2e27255 | 2347 | if (timeout) { |
766b9f92 DD |
2348 | ktime_get_ts64(&timeout64); |
2349 | *timeout = timespec64_to_timespec( | |
2350 | timespec64_sub(end_time, timeout64)); | |
a2e27255 ACM |
2351 | if (timeout->tv_sec < 0) { |
2352 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2353 | break; | |
2354 | } | |
2355 | ||
2356 | /* Timeout, return less than vlen datagrams */ | |
2357 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2358 | break; | |
2359 | } | |
2360 | ||
2361 | /* Out of band data, return right away */ | |
2362 | if (msg_sys.msg_flags & MSG_OOB) | |
2363 | break; | |
a78cb84c | 2364 | cond_resched(); |
a2e27255 ACM |
2365 | } |
2366 | ||
a2e27255 | 2367 | if (err == 0) |
34b88a68 ACM |
2368 | goto out_put; |
2369 | ||
2370 | if (datagrams == 0) { | |
2371 | datagrams = err; | |
2372 | goto out_put; | |
2373 | } | |
a2e27255 | 2374 | |
34b88a68 ACM |
2375 | /* |
2376 | * We may return less entries than requested (vlen) if the | |
2377 | * sock is non block and there aren't enough datagrams... | |
2378 | */ | |
2379 | if (err != -EAGAIN) { | |
a2e27255 | 2380 | /* |
34b88a68 ACM |
2381 | * ... or if recvmsg returns an error after we |
2382 | * received some datagrams, where we record the | |
2383 | * error to return on the next call or if the | |
2384 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2385 | */ |
34b88a68 | 2386 | sock->sk->sk_err = -err; |
a2e27255 | 2387 | } |
34b88a68 ACM |
2388 | out_put: |
2389 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2390 | |
34b88a68 | 2391 | return datagrams; |
a2e27255 ACM |
2392 | } |
2393 | ||
2394 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2395 | unsigned int, vlen, unsigned int, flags, | |
2396 | struct timespec __user *, timeout) | |
2397 | { | |
2398 | int datagrams; | |
2399 | struct timespec timeout_sys; | |
2400 | ||
1be374a0 AL |
2401 | if (flags & MSG_CMSG_COMPAT) |
2402 | return -EINVAL; | |
2403 | ||
a2e27255 ACM |
2404 | if (!timeout) |
2405 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2406 | ||
2407 | if (copy_from_user(&timeout_sys, timeout, sizeof(timeout_sys))) | |
2408 | return -EFAULT; | |
2409 | ||
2410 | datagrams = __sys_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
2411 | ||
2412 | if (datagrams > 0 && | |
2413 | copy_to_user(timeout, &timeout_sys, sizeof(timeout_sys))) | |
2414 | datagrams = -EFAULT; | |
2415 | ||
2416 | return datagrams; | |
2417 | } | |
2418 | ||
2419 | #ifdef __ARCH_WANT_SYS_SOCKETCALL | |
1da177e4 LT |
2420 | /* Argument list sizes for sys_socketcall */ |
2421 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2422 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2423 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2424 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2425 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2426 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2427 | }; |
2428 | ||
1da177e4 LT |
2429 | #undef AL |
2430 | ||
2431 | /* | |
89bddce5 | 2432 | * System call vectors. |
1da177e4 LT |
2433 | * |
2434 | * Argument checking cleaned up. Saved 20% in size. | |
2435 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2436 | * it is set by the callees. |
1da177e4 LT |
2437 | */ |
2438 | ||
3e0fa65f | 2439 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2440 | { |
2950fa9d | 2441 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2442 | unsigned long a0, a1; |
1da177e4 | 2443 | int err; |
47379052 | 2444 | unsigned int len; |
1da177e4 | 2445 | |
228e548e | 2446 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 | 2447 | return -EINVAL; |
45c8178c | 2448 | call = array_index_nospec(call, SYS_SENDMMSG + 1); |
1da177e4 | 2449 | |
47379052 AV |
2450 | len = nargs[call]; |
2451 | if (len > sizeof(a)) | |
2452 | return -EINVAL; | |
2453 | ||
1da177e4 | 2454 | /* copy_from_user should be SMP safe. */ |
47379052 | 2455 | if (copy_from_user(a, args, len)) |
1da177e4 | 2456 | return -EFAULT; |
3ec3b2fb | 2457 | |
2950fa9d CG |
2458 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2459 | if (err) | |
2460 | return err; | |
3ec3b2fb | 2461 | |
89bddce5 SH |
2462 | a0 = a[0]; |
2463 | a1 = a[1]; | |
2464 | ||
2465 | switch (call) { | |
2466 | case SYS_SOCKET: | |
2467 | err = sys_socket(a0, a1, a[2]); | |
2468 | break; | |
2469 | case SYS_BIND: | |
2470 | err = sys_bind(a0, (struct sockaddr __user *)a1, a[2]); | |
2471 | break; | |
2472 | case SYS_CONNECT: | |
2473 | err = sys_connect(a0, (struct sockaddr __user *)a1, a[2]); | |
2474 | break; | |
2475 | case SYS_LISTEN: | |
2476 | err = sys_listen(a0, a1); | |
2477 | break; | |
2478 | case SYS_ACCEPT: | |
de11defe UD |
2479 | err = sys_accept4(a0, (struct sockaddr __user *)a1, |
2480 | (int __user *)a[2], 0); | |
89bddce5 SH |
2481 | break; |
2482 | case SYS_GETSOCKNAME: | |
2483 | err = | |
2484 | sys_getsockname(a0, (struct sockaddr __user *)a1, | |
2485 | (int __user *)a[2]); | |
2486 | break; | |
2487 | case SYS_GETPEERNAME: | |
2488 | err = | |
2489 | sys_getpeername(a0, (struct sockaddr __user *)a1, | |
2490 | (int __user *)a[2]); | |
2491 | break; | |
2492 | case SYS_SOCKETPAIR: | |
2493 | err = sys_socketpair(a0, a1, a[2], (int __user *)a[3]); | |
2494 | break; | |
2495 | case SYS_SEND: | |
2496 | err = sys_send(a0, (void __user *)a1, a[2], a[3]); | |
2497 | break; | |
2498 | case SYS_SENDTO: | |
2499 | err = sys_sendto(a0, (void __user *)a1, a[2], a[3], | |
2500 | (struct sockaddr __user *)a[4], a[5]); | |
2501 | break; | |
2502 | case SYS_RECV: | |
2503 | err = sys_recv(a0, (void __user *)a1, a[2], a[3]); | |
2504 | break; | |
2505 | case SYS_RECVFROM: | |
2506 | err = sys_recvfrom(a0, (void __user *)a1, a[2], a[3], | |
2507 | (struct sockaddr __user *)a[4], | |
2508 | (int __user *)a[5]); | |
2509 | break; | |
2510 | case SYS_SHUTDOWN: | |
2511 | err = sys_shutdown(a0, a1); | |
2512 | break; | |
2513 | case SYS_SETSOCKOPT: | |
2514 | err = sys_setsockopt(a0, a1, a[2], (char __user *)a[3], a[4]); | |
2515 | break; | |
2516 | case SYS_GETSOCKOPT: | |
2517 | err = | |
2518 | sys_getsockopt(a0, a1, a[2], (char __user *)a[3], | |
2519 | (int __user *)a[4]); | |
2520 | break; | |
2521 | case SYS_SENDMSG: | |
666547ff | 2522 | err = sys_sendmsg(a0, (struct user_msghdr __user *)a1, a[2]); |
89bddce5 | 2523 | break; |
228e548e AB |
2524 | case SYS_SENDMMSG: |
2525 | err = sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3]); | |
2526 | break; | |
89bddce5 | 2527 | case SYS_RECVMSG: |
666547ff | 2528 | err = sys_recvmsg(a0, (struct user_msghdr __user *)a1, a[2]); |
89bddce5 | 2529 | break; |
a2e27255 ACM |
2530 | case SYS_RECVMMSG: |
2531 | err = sys_recvmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3], | |
2532 | (struct timespec __user *)a[4]); | |
2533 | break; | |
de11defe UD |
2534 | case SYS_ACCEPT4: |
2535 | err = sys_accept4(a0, (struct sockaddr __user *)a1, | |
2536 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2537 | break; |
89bddce5 SH |
2538 | default: |
2539 | err = -EINVAL; | |
2540 | break; | |
1da177e4 LT |
2541 | } |
2542 | return err; | |
2543 | } | |
2544 | ||
89bddce5 | 2545 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2546 | |
55737fda SH |
2547 | /** |
2548 | * sock_register - add a socket protocol handler | |
2549 | * @ops: description of protocol | |
2550 | * | |
1da177e4 LT |
2551 | * This function is called by a protocol handler that wants to |
2552 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2553 | * socket interface. The value ops->family corresponds to the |
55737fda | 2554 | * socket system call protocol family. |
1da177e4 | 2555 | */ |
f0fd27d4 | 2556 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2557 | { |
2558 | int err; | |
2559 | ||
2560 | if (ops->family >= NPROTO) { | |
3410f22e | 2561 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2562 | return -ENOBUFS; |
2563 | } | |
55737fda SH |
2564 | |
2565 | spin_lock(&net_family_lock); | |
190683a9 ED |
2566 | if (rcu_dereference_protected(net_families[ops->family], |
2567 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2568 | err = -EEXIST; |
2569 | else { | |
cf778b00 | 2570 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2571 | err = 0; |
2572 | } | |
55737fda SH |
2573 | spin_unlock(&net_family_lock); |
2574 | ||
3410f22e | 2575 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2576 | return err; |
2577 | } | |
c6d409cf | 2578 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2579 | |
55737fda SH |
2580 | /** |
2581 | * sock_unregister - remove a protocol handler | |
2582 | * @family: protocol family to remove | |
2583 | * | |
1da177e4 LT |
2584 | * This function is called by a protocol handler that wants to |
2585 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2586 | * new socket creation. |
2587 | * | |
2588 | * If protocol handler is a module, then it can use module reference | |
2589 | * counts to protect against new references. If protocol handler is not | |
2590 | * a module then it needs to provide its own protection in | |
2591 | * the ops->create routine. | |
1da177e4 | 2592 | */ |
f0fd27d4 | 2593 | void sock_unregister(int family) |
1da177e4 | 2594 | { |
f0fd27d4 | 2595 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2596 | |
55737fda | 2597 | spin_lock(&net_family_lock); |
a9b3cd7f | 2598 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2599 | spin_unlock(&net_family_lock); |
2600 | ||
2601 | synchronize_rcu(); | |
2602 | ||
3410f22e | 2603 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2604 | } |
c6d409cf | 2605 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2606 | |
77d76ea3 | 2607 | static int __init sock_init(void) |
1da177e4 | 2608 | { |
b3e19d92 | 2609 | int err; |
2ca794e5 EB |
2610 | /* |
2611 | * Initialize the network sysctl infrastructure. | |
2612 | */ | |
2613 | err = net_sysctl_init(); | |
2614 | if (err) | |
2615 | goto out; | |
b3e19d92 | 2616 | |
1da177e4 | 2617 | /* |
89bddce5 | 2618 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2619 | */ |
2620 | skb_init(); | |
1da177e4 LT |
2621 | |
2622 | /* | |
89bddce5 | 2623 | * Initialize the protocols module. |
1da177e4 LT |
2624 | */ |
2625 | ||
2626 | init_inodecache(); | |
b3e19d92 NP |
2627 | |
2628 | err = register_filesystem(&sock_fs_type); | |
2629 | if (err) | |
2630 | goto out_fs; | |
1da177e4 | 2631 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2632 | if (IS_ERR(sock_mnt)) { |
2633 | err = PTR_ERR(sock_mnt); | |
2634 | goto out_mount; | |
2635 | } | |
77d76ea3 AK |
2636 | |
2637 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2638 | */ |
2639 | ||
2640 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
2641 | err = netfilter_init(); |
2642 | if (err) | |
2643 | goto out; | |
1da177e4 | 2644 | #endif |
cbeb321a | 2645 | |
408eccce | 2646 | ptp_classifier_init(); |
c1f19b51 | 2647 | |
b3e19d92 NP |
2648 | out: |
2649 | return err; | |
2650 | ||
2651 | out_mount: | |
2652 | unregister_filesystem(&sock_fs_type); | |
2653 | out_fs: | |
2654 | goto out; | |
1da177e4 LT |
2655 | } |
2656 | ||
77d76ea3 AK |
2657 | core_initcall(sock_init); /* early initcall */ |
2658 | ||
6fde36d5 AS |
2659 | static int __init jit_init(void) |
2660 | { | |
2661 | #ifdef CONFIG_BPF_JIT_ALWAYS_ON | |
2662 | bpf_jit_enable = 1; | |
2663 | #endif | |
2664 | return 0; | |
2665 | } | |
2666 | pure_initcall(jit_init); | |
2667 | ||
1da177e4 LT |
2668 | #ifdef CONFIG_PROC_FS |
2669 | void socket_seq_show(struct seq_file *seq) | |
2670 | { | |
2671 | int cpu; | |
2672 | int counter = 0; | |
2673 | ||
6f912042 | 2674 | for_each_possible_cpu(cpu) |
89bddce5 | 2675 | counter += per_cpu(sockets_in_use, cpu); |
1da177e4 LT |
2676 | |
2677 | /* It can be negative, by the way. 8) */ | |
2678 | if (counter < 0) | |
2679 | counter = 0; | |
2680 | ||
2681 | seq_printf(seq, "sockets: used %d\n", counter); | |
2682 | } | |
89bddce5 | 2683 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 2684 | |
89bbfc95 | 2685 | #ifdef CONFIG_COMPAT |
6b96018b | 2686 | static int do_siocgstamp(struct net *net, struct socket *sock, |
644595f8 | 2687 | unsigned int cmd, void __user *up) |
7a229387 | 2688 | { |
7a229387 AB |
2689 | mm_segment_t old_fs = get_fs(); |
2690 | struct timeval ktv; | |
2691 | int err; | |
2692 | ||
2693 | set_fs(KERNEL_DS); | |
6b96018b | 2694 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&ktv); |
7a229387 | 2695 | set_fs(old_fs); |
644595f8 | 2696 | if (!err) |
ed6fe9d6 | 2697 | err = compat_put_timeval(&ktv, up); |
644595f8 | 2698 | |
7a229387 AB |
2699 | return err; |
2700 | } | |
2701 | ||
6b96018b | 2702 | static int do_siocgstampns(struct net *net, struct socket *sock, |
644595f8 | 2703 | unsigned int cmd, void __user *up) |
7a229387 | 2704 | { |
7a229387 AB |
2705 | mm_segment_t old_fs = get_fs(); |
2706 | struct timespec kts; | |
2707 | int err; | |
2708 | ||
2709 | set_fs(KERNEL_DS); | |
6b96018b | 2710 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&kts); |
7a229387 | 2711 | set_fs(old_fs); |
644595f8 | 2712 | if (!err) |
ed6fe9d6 | 2713 | err = compat_put_timespec(&kts, up); |
644595f8 | 2714 | |
7a229387 AB |
2715 | return err; |
2716 | } | |
2717 | ||
6b96018b | 2718 | static int dev_ifname32(struct net *net, struct compat_ifreq __user *uifr32) |
7a229387 AB |
2719 | { |
2720 | struct ifreq __user *uifr; | |
2721 | int err; | |
2722 | ||
2723 | uifr = compat_alloc_user_space(sizeof(struct ifreq)); | |
6b96018b | 2724 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2725 | return -EFAULT; |
2726 | ||
6b96018b | 2727 | err = dev_ioctl(net, SIOCGIFNAME, uifr); |
7a229387 AB |
2728 | if (err) |
2729 | return err; | |
2730 | ||
6b96018b | 2731 | if (copy_in_user(uifr32, uifr, sizeof(struct compat_ifreq))) |
7a229387 AB |
2732 | return -EFAULT; |
2733 | ||
2734 | return 0; | |
2735 | } | |
2736 | ||
6b96018b | 2737 | static int dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 2738 | { |
6b96018b | 2739 | struct compat_ifconf ifc32; |
7a229387 AB |
2740 | struct ifconf ifc; |
2741 | struct ifconf __user *uifc; | |
6b96018b | 2742 | struct compat_ifreq __user *ifr32; |
7a229387 AB |
2743 | struct ifreq __user *ifr; |
2744 | unsigned int i, j; | |
2745 | int err; | |
2746 | ||
6b96018b | 2747 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2748 | return -EFAULT; |
2749 | ||
43da5f2e | 2750 | memset(&ifc, 0, sizeof(ifc)); |
7a229387 AB |
2751 | if (ifc32.ifcbuf == 0) { |
2752 | ifc32.ifc_len = 0; | |
2753 | ifc.ifc_len = 0; | |
2754 | ifc.ifc_req = NULL; | |
2755 | uifc = compat_alloc_user_space(sizeof(struct ifconf)); | |
2756 | } else { | |
c6d409cf ED |
2757 | size_t len = ((ifc32.ifc_len / sizeof(struct compat_ifreq)) + 1) * |
2758 | sizeof(struct ifreq); | |
7a229387 AB |
2759 | uifc = compat_alloc_user_space(sizeof(struct ifconf) + len); |
2760 | ifc.ifc_len = len; | |
2761 | ifr = ifc.ifc_req = (void __user *)(uifc + 1); | |
2762 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
c6d409cf | 2763 | for (i = 0; i < ifc32.ifc_len; i += sizeof(struct compat_ifreq)) { |
6b96018b | 2764 | if (copy_in_user(ifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2765 | return -EFAULT; |
2766 | ifr++; | |
2767 | ifr32++; | |
2768 | } | |
2769 | } | |
2770 | if (copy_to_user(uifc, &ifc, sizeof(struct ifconf))) | |
2771 | return -EFAULT; | |
2772 | ||
6b96018b | 2773 | err = dev_ioctl(net, SIOCGIFCONF, uifc); |
7a229387 AB |
2774 | if (err) |
2775 | return err; | |
2776 | ||
2777 | if (copy_from_user(&ifc, uifc, sizeof(struct ifconf))) | |
2778 | return -EFAULT; | |
2779 | ||
2780 | ifr = ifc.ifc_req; | |
2781 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
2782 | for (i = 0, j = 0; | |
c6d409cf ED |
2783 | i + sizeof(struct compat_ifreq) <= ifc32.ifc_len && j < ifc.ifc_len; |
2784 | i += sizeof(struct compat_ifreq), j += sizeof(struct ifreq)) { | |
2785 | if (copy_in_user(ifr32, ifr, sizeof(struct compat_ifreq))) | |
7a229387 AB |
2786 | return -EFAULT; |
2787 | ifr32++; | |
2788 | ifr++; | |
2789 | } | |
2790 | ||
2791 | if (ifc32.ifcbuf == 0) { | |
2792 | /* Translate from 64-bit structure multiple to | |
2793 | * a 32-bit one. | |
2794 | */ | |
2795 | i = ifc.ifc_len; | |
6b96018b | 2796 | i = ((i / sizeof(struct ifreq)) * sizeof(struct compat_ifreq)); |
7a229387 AB |
2797 | ifc32.ifc_len = i; |
2798 | } else { | |
2799 | ifc32.ifc_len = i; | |
2800 | } | |
6b96018b | 2801 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2802 | return -EFAULT; |
2803 | ||
2804 | return 0; | |
2805 | } | |
2806 | ||
6b96018b | 2807 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 2808 | { |
3a7da39d BH |
2809 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
2810 | bool convert_in = false, convert_out = false; | |
2811 | size_t buf_size = ALIGN(sizeof(struct ifreq), 8); | |
2812 | struct ethtool_rxnfc __user *rxnfc; | |
7a229387 | 2813 | struct ifreq __user *ifr; |
3a7da39d BH |
2814 | u32 rule_cnt = 0, actual_rule_cnt; |
2815 | u32 ethcmd; | |
7a229387 | 2816 | u32 data; |
3a7da39d | 2817 | int ret; |
7a229387 | 2818 | |
3a7da39d BH |
2819 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
2820 | return -EFAULT; | |
7a229387 | 2821 | |
3a7da39d BH |
2822 | compat_rxnfc = compat_ptr(data); |
2823 | ||
2824 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
2825 | return -EFAULT; |
2826 | ||
3a7da39d BH |
2827 | /* Most ethtool structures are defined without padding. |
2828 | * Unfortunately struct ethtool_rxnfc is an exception. | |
2829 | */ | |
2830 | switch (ethcmd) { | |
2831 | default: | |
2832 | break; | |
2833 | case ETHTOOL_GRXCLSRLALL: | |
2834 | /* Buffer size is variable */ | |
2835 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
2836 | return -EFAULT; | |
2837 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
2838 | return -ENOMEM; | |
2839 | buf_size += rule_cnt * sizeof(u32); | |
2840 | /* fall through */ | |
2841 | case ETHTOOL_GRXRINGS: | |
2842 | case ETHTOOL_GRXCLSRLCNT: | |
2843 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 2844 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
2845 | convert_out = true; |
2846 | /* fall through */ | |
2847 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
2848 | buf_size += sizeof(struct ethtool_rxnfc); |
2849 | convert_in = true; | |
2850 | break; | |
2851 | } | |
2852 | ||
2853 | ifr = compat_alloc_user_space(buf_size); | |
954b1244 | 2854 | rxnfc = (void __user *)ifr + ALIGN(sizeof(struct ifreq), 8); |
3a7da39d BH |
2855 | |
2856 | if (copy_in_user(&ifr->ifr_name, &ifr32->ifr_name, IFNAMSIZ)) | |
7a229387 AB |
2857 | return -EFAULT; |
2858 | ||
3a7da39d BH |
2859 | if (put_user(convert_in ? rxnfc : compat_ptr(data), |
2860 | &ifr->ifr_ifru.ifru_data)) | |
7a229387 AB |
2861 | return -EFAULT; |
2862 | ||
3a7da39d | 2863 | if (convert_in) { |
127fe533 | 2864 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
2865 | * fs.ring_cookie and at the end of fs, but nowhere else. |
2866 | */ | |
127fe533 AD |
2867 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
2868 | sizeof(compat_rxnfc->fs.m_ext) != | |
2869 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
2870 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
2871 | BUILD_BUG_ON( |
2872 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
2873 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
2874 | offsetof(struct ethtool_rxnfc, fs.location) - | |
2875 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
2876 | ||
2877 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
2878 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
2879 | (void __user *)rxnfc) || | |
3a7da39d BH |
2880 | copy_in_user(&rxnfc->fs.ring_cookie, |
2881 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 | 2882 | (void __user *)(&rxnfc->fs.location + 1) - |
7d584568 WW |
2883 | (void __user *)&rxnfc->fs.ring_cookie)) |
2884 | return -EFAULT; | |
2885 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
2886 | if (put_user(rule_cnt, &rxnfc->rule_cnt)) | |
2887 | return -EFAULT; | |
2888 | } else if (copy_in_user(&rxnfc->rule_cnt, | |
2889 | &compat_rxnfc->rule_cnt, | |
2890 | sizeof(rxnfc->rule_cnt))) | |
3a7da39d BH |
2891 | return -EFAULT; |
2892 | } | |
2893 | ||
2894 | ret = dev_ioctl(net, SIOCETHTOOL, ifr); | |
2895 | if (ret) | |
2896 | return ret; | |
2897 | ||
2898 | if (convert_out) { | |
2899 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
2900 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
2901 | (const void __user *)rxnfc) || | |
3a7da39d BH |
2902 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
2903 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
2904 | (const void __user *)(&rxnfc->fs.location + 1) - |
2905 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2906 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
2907 | sizeof(rxnfc->rule_cnt))) | |
2908 | return -EFAULT; | |
2909 | ||
2910 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
2911 | /* As an optimisation, we only copy the actual | |
2912 | * number of rules that the underlying | |
2913 | * function returned. Since Mallory might | |
2914 | * change the rule count in user memory, we | |
2915 | * check that it is less than the rule count | |
2916 | * originally given (as the user buffer size), | |
2917 | * which has been range-checked. | |
2918 | */ | |
2919 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
2920 | return -EFAULT; | |
2921 | if (actual_rule_cnt < rule_cnt) | |
2922 | rule_cnt = actual_rule_cnt; | |
2923 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
2924 | &rxnfc->rule_locs[0], | |
2925 | rule_cnt * sizeof(u32))) | |
2926 | return -EFAULT; | |
2927 | } | |
2928 | } | |
2929 | ||
2930 | return 0; | |
7a229387 AB |
2931 | } |
2932 | ||
7a50a240 AB |
2933 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
2934 | { | |
2935 | void __user *uptr; | |
2936 | compat_uptr_t uptr32; | |
2937 | struct ifreq __user *uifr; | |
2938 | ||
c6d409cf | 2939 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
7a50a240 AB |
2940 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
2941 | return -EFAULT; | |
2942 | ||
2943 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
2944 | return -EFAULT; | |
2945 | ||
2946 | uptr = compat_ptr(uptr32); | |
2947 | ||
2948 | if (put_user(uptr, &uifr->ifr_settings.ifs_ifsu.raw_hdlc)) | |
2949 | return -EFAULT; | |
2950 | ||
2951 | return dev_ioctl(net, SIOCWANDEV, uifr); | |
2952 | } | |
2953 | ||
6b96018b AB |
2954 | static int bond_ioctl(struct net *net, unsigned int cmd, |
2955 | struct compat_ifreq __user *ifr32) | |
7a229387 AB |
2956 | { |
2957 | struct ifreq kifr; | |
7a229387 AB |
2958 | mm_segment_t old_fs; |
2959 | int err; | |
7a229387 AB |
2960 | |
2961 | switch (cmd) { | |
2962 | case SIOCBONDENSLAVE: | |
2963 | case SIOCBONDRELEASE: | |
2964 | case SIOCBONDSETHWADDR: | |
2965 | case SIOCBONDCHANGEACTIVE: | |
6b96018b | 2966 | if (copy_from_user(&kifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2967 | return -EFAULT; |
2968 | ||
2969 | old_fs = get_fs(); | |
c6d409cf | 2970 | set_fs(KERNEL_DS); |
c3f52ae6 | 2971 | err = dev_ioctl(net, cmd, |
2972 | (struct ifreq __user __force *) &kifr); | |
c6d409cf | 2973 | set_fs(old_fs); |
7a229387 AB |
2974 | |
2975 | return err; | |
7a229387 | 2976 | default: |
07d106d0 | 2977 | return -ENOIOCTLCMD; |
ccbd6a5a | 2978 | } |
7a229387 AB |
2979 | } |
2980 | ||
590d4693 BH |
2981 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
2982 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 2983 | struct compat_ifreq __user *u_ifreq32) |
7a229387 AB |
2984 | { |
2985 | struct ifreq __user *u_ifreq64; | |
7a229387 AB |
2986 | char tmp_buf[IFNAMSIZ]; |
2987 | void __user *data64; | |
2988 | u32 data32; | |
2989 | ||
2990 | if (copy_from_user(&tmp_buf[0], &(u_ifreq32->ifr_ifrn.ifrn_name[0]), | |
2991 | IFNAMSIZ)) | |
2992 | return -EFAULT; | |
417c3522 | 2993 | if (get_user(data32, &u_ifreq32->ifr_ifru.ifru_data)) |
7a229387 AB |
2994 | return -EFAULT; |
2995 | data64 = compat_ptr(data32); | |
2996 | ||
2997 | u_ifreq64 = compat_alloc_user_space(sizeof(*u_ifreq64)); | |
2998 | ||
7a229387 AB |
2999 | if (copy_to_user(&u_ifreq64->ifr_ifrn.ifrn_name[0], &tmp_buf[0], |
3000 | IFNAMSIZ)) | |
3001 | return -EFAULT; | |
417c3522 | 3002 | if (put_user(data64, &u_ifreq64->ifr_ifru.ifru_data)) |
7a229387 AB |
3003 | return -EFAULT; |
3004 | ||
6b96018b | 3005 | return dev_ioctl(net, cmd, u_ifreq64); |
7a229387 AB |
3006 | } |
3007 | ||
6b96018b AB |
3008 | static int dev_ifsioc(struct net *net, struct socket *sock, |
3009 | unsigned int cmd, struct compat_ifreq __user *uifr32) | |
7a229387 | 3010 | { |
a2116ed2 | 3011 | struct ifreq __user *uifr; |
7a229387 AB |
3012 | int err; |
3013 | ||
a2116ed2 AB |
3014 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
3015 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
3016 | return -EFAULT; | |
3017 | ||
3018 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
3019 | ||
7a229387 AB |
3020 | if (!err) { |
3021 | switch (cmd) { | |
3022 | case SIOCGIFFLAGS: | |
3023 | case SIOCGIFMETRIC: | |
3024 | case SIOCGIFMTU: | |
3025 | case SIOCGIFMEM: | |
3026 | case SIOCGIFHWADDR: | |
3027 | case SIOCGIFINDEX: | |
3028 | case SIOCGIFADDR: | |
3029 | case SIOCGIFBRDADDR: | |
3030 | case SIOCGIFDSTADDR: | |
3031 | case SIOCGIFNETMASK: | |
fab2532b | 3032 | case SIOCGIFPFLAGS: |
7a229387 | 3033 | case SIOCGIFTXQLEN: |
fab2532b AB |
3034 | case SIOCGMIIPHY: |
3035 | case SIOCGMIIREG: | |
a2116ed2 | 3036 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
7a229387 AB |
3037 | err = -EFAULT; |
3038 | break; | |
3039 | } | |
3040 | } | |
3041 | return err; | |
3042 | } | |
3043 | ||
a2116ed2 AB |
3044 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
3045 | struct compat_ifreq __user *uifr32) | |
3046 | { | |
3047 | struct ifreq ifr; | |
3048 | struct compat_ifmap __user *uifmap32; | |
3049 | mm_segment_t old_fs; | |
3050 | int err; | |
3051 | ||
3052 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
3053 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3054 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3055 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3056 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3057 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3058 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3059 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3060 | if (err) |
3061 | return -EFAULT; | |
3062 | ||
3063 | old_fs = get_fs(); | |
c6d409cf | 3064 | set_fs(KERNEL_DS); |
c3f52ae6 | 3065 | err = dev_ioctl(net, cmd, (void __user __force *)&ifr); |
c6d409cf | 3066 | set_fs(old_fs); |
a2116ed2 AB |
3067 | |
3068 | if (cmd == SIOCGIFMAP && !err) { | |
3069 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3070 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3071 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3072 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3073 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3074 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3075 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3076 | if (err) |
3077 | err = -EFAULT; | |
3078 | } | |
3079 | return err; | |
3080 | } | |
3081 | ||
7a229387 | 3082 | struct rtentry32 { |
c6d409cf | 3083 | u32 rt_pad1; |
7a229387 AB |
3084 | struct sockaddr rt_dst; /* target address */ |
3085 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
3086 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
3087 | unsigned short rt_flags; |
3088 | short rt_pad2; | |
3089 | u32 rt_pad3; | |
3090 | unsigned char rt_tos; | |
3091 | unsigned char rt_class; | |
3092 | short rt_pad4; | |
3093 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 3094 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
3095 | u32 rt_mtu; /* per route MTU/Window */ |
3096 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
3097 | unsigned short rt_irtt; /* Initial RTT */ |
3098 | }; | |
3099 | ||
3100 | struct in6_rtmsg32 { | |
3101 | struct in6_addr rtmsg_dst; | |
3102 | struct in6_addr rtmsg_src; | |
3103 | struct in6_addr rtmsg_gateway; | |
3104 | u32 rtmsg_type; | |
3105 | u16 rtmsg_dst_len; | |
3106 | u16 rtmsg_src_len; | |
3107 | u32 rtmsg_metric; | |
3108 | u32 rtmsg_info; | |
3109 | u32 rtmsg_flags; | |
3110 | s32 rtmsg_ifindex; | |
3111 | }; | |
3112 | ||
6b96018b AB |
3113 | static int routing_ioctl(struct net *net, struct socket *sock, |
3114 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3115 | { |
3116 | int ret; | |
3117 | void *r = NULL; | |
3118 | struct in6_rtmsg r6; | |
3119 | struct rtentry r4; | |
3120 | char devname[16]; | |
3121 | u32 rtdev; | |
3122 | mm_segment_t old_fs = get_fs(); | |
3123 | ||
6b96018b AB |
3124 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3125 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3126 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3127 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
3128 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3129 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3130 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3131 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3132 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3133 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3134 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3135 | |
3136 | r = (void *) &r6; | |
3137 | } else { /* ipv4 */ | |
6b96018b | 3138 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3139 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3140 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
3141 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
3142 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3143 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3144 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
3145 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3146 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3147 | if (rtdev) { |
c6d409cf | 3148 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3149 | r4.rt_dev = (char __user __force *)devname; |
3150 | devname[15] = 0; | |
7a229387 AB |
3151 | } else |
3152 | r4.rt_dev = NULL; | |
3153 | ||
3154 | r = (void *) &r4; | |
3155 | } | |
3156 | ||
3157 | if (ret) { | |
3158 | ret = -EFAULT; | |
3159 | goto out; | |
3160 | } | |
3161 | ||
c6d409cf | 3162 | set_fs(KERNEL_DS); |
6b96018b | 3163 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3164 | set_fs(old_fs); |
7a229387 AB |
3165 | |
3166 | out: | |
7a229387 AB |
3167 | return ret; |
3168 | } | |
3169 | ||
3170 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3171 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3172 | * use compatible ioctls |
7a229387 | 3173 | */ |
6b96018b | 3174 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3175 | { |
6b96018b | 3176 | compat_ulong_t tmp; |
7a229387 | 3177 | |
6b96018b | 3178 | if (get_user(tmp, argp)) |
7a229387 AB |
3179 | return -EFAULT; |
3180 | if (tmp == BRCTL_GET_VERSION) | |
3181 | return BRCTL_VERSION + 1; | |
3182 | return -EINVAL; | |
3183 | } | |
3184 | ||
6b96018b AB |
3185 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3186 | unsigned int cmd, unsigned long arg) | |
3187 | { | |
3188 | void __user *argp = compat_ptr(arg); | |
3189 | struct sock *sk = sock->sk; | |
3190 | struct net *net = sock_net(sk); | |
7a229387 | 3191 | |
6b96018b | 3192 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3193 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3194 | |
3195 | switch (cmd) { | |
3196 | case SIOCSIFBR: | |
3197 | case SIOCGIFBR: | |
3198 | return old_bridge_ioctl(argp); | |
3199 | case SIOCGIFNAME: | |
3200 | return dev_ifname32(net, argp); | |
3201 | case SIOCGIFCONF: | |
3202 | return dev_ifconf(net, argp); | |
3203 | case SIOCETHTOOL: | |
3204 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3205 | case SIOCWANDEV: |
3206 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3207 | case SIOCGIFMAP: |
3208 | case SIOCSIFMAP: | |
3209 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3210 | case SIOCBONDENSLAVE: |
3211 | case SIOCBONDRELEASE: | |
3212 | case SIOCBONDSETHWADDR: | |
6b96018b AB |
3213 | case SIOCBONDCHANGEACTIVE: |
3214 | return bond_ioctl(net, cmd, argp); | |
3215 | case SIOCADDRT: | |
3216 | case SIOCDELRT: | |
3217 | return routing_ioctl(net, sock, cmd, argp); | |
3218 | case SIOCGSTAMP: | |
3219 | return do_siocgstamp(net, sock, cmd, argp); | |
3220 | case SIOCGSTAMPNS: | |
3221 | return do_siocgstampns(net, sock, cmd, argp); | |
590d4693 BH |
3222 | case SIOCBONDSLAVEINFOQUERY: |
3223 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3224 | case SIOCSHWTSTAMP: |
fd468c74 | 3225 | case SIOCGHWTSTAMP: |
590d4693 | 3226 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3227 | |
3228 | case FIOSETOWN: | |
3229 | case SIOCSPGRP: | |
3230 | case FIOGETOWN: | |
3231 | case SIOCGPGRP: | |
3232 | case SIOCBRADDBR: | |
3233 | case SIOCBRDELBR: | |
3234 | case SIOCGIFVLAN: | |
3235 | case SIOCSIFVLAN: | |
3236 | case SIOCADDDLCI: | |
3237 | case SIOCDELDLCI: | |
c62cce2c | 3238 | case SIOCGSKNS: |
6b96018b AB |
3239 | return sock_ioctl(file, cmd, arg); |
3240 | ||
3241 | case SIOCGIFFLAGS: | |
3242 | case SIOCSIFFLAGS: | |
3243 | case SIOCGIFMETRIC: | |
3244 | case SIOCSIFMETRIC: | |
3245 | case SIOCGIFMTU: | |
3246 | case SIOCSIFMTU: | |
3247 | case SIOCGIFMEM: | |
3248 | case SIOCSIFMEM: | |
3249 | case SIOCGIFHWADDR: | |
3250 | case SIOCSIFHWADDR: | |
3251 | case SIOCADDMULTI: | |
3252 | case SIOCDELMULTI: | |
3253 | case SIOCGIFINDEX: | |
6b96018b AB |
3254 | case SIOCGIFADDR: |
3255 | case SIOCSIFADDR: | |
3256 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3257 | case SIOCDIFADDR: |
6b96018b AB |
3258 | case SIOCGIFBRDADDR: |
3259 | case SIOCSIFBRDADDR: | |
3260 | case SIOCGIFDSTADDR: | |
3261 | case SIOCSIFDSTADDR: | |
3262 | case SIOCGIFNETMASK: | |
3263 | case SIOCSIFNETMASK: | |
3264 | case SIOCSIFPFLAGS: | |
3265 | case SIOCGIFPFLAGS: | |
3266 | case SIOCGIFTXQLEN: | |
3267 | case SIOCSIFTXQLEN: | |
3268 | case SIOCBRADDIF: | |
3269 | case SIOCBRDELIF: | |
9177efd3 AB |
3270 | case SIOCSIFNAME: |
3271 | case SIOCGMIIPHY: | |
3272 | case SIOCGMIIREG: | |
3273 | case SIOCSMIIREG: | |
6b96018b | 3274 | return dev_ifsioc(net, sock, cmd, argp); |
9177efd3 | 3275 | |
6b96018b AB |
3276 | case SIOCSARP: |
3277 | case SIOCGARP: | |
3278 | case SIOCDARP: | |
6b96018b | 3279 | case SIOCATMARK: |
9177efd3 AB |
3280 | return sock_do_ioctl(net, sock, cmd, arg); |
3281 | } | |
3282 | ||
6b96018b AB |
3283 | return -ENOIOCTLCMD; |
3284 | } | |
7a229387 | 3285 | |
95c96174 | 3286 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3287 | unsigned long arg) |
89bbfc95 SP |
3288 | { |
3289 | struct socket *sock = file->private_data; | |
3290 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3291 | struct sock *sk; |
3292 | struct net *net; | |
3293 | ||
3294 | sk = sock->sk; | |
3295 | net = sock_net(sk); | |
89bbfc95 SP |
3296 | |
3297 | if (sock->ops->compat_ioctl) | |
3298 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3299 | ||
87de87d5 DM |
3300 | if (ret == -ENOIOCTLCMD && |
3301 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3302 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3303 | ||
6b96018b AB |
3304 | if (ret == -ENOIOCTLCMD) |
3305 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3306 | ||
89bbfc95 SP |
3307 | return ret; |
3308 | } | |
3309 | #endif | |
3310 | ||
ac5a488e SS |
3311 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3312 | { | |
3313 | return sock->ops->bind(sock, addr, addrlen); | |
3314 | } | |
c6d409cf | 3315 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e SS |
3316 | |
3317 | int kernel_listen(struct socket *sock, int backlog) | |
3318 | { | |
3319 | return sock->ops->listen(sock, backlog); | |
3320 | } | |
c6d409cf | 3321 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e SS |
3322 | |
3323 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) | |
3324 | { | |
3325 | struct sock *sk = sock->sk; | |
3326 | int err; | |
3327 | ||
3328 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3329 | newsock); | |
3330 | if (err < 0) | |
3331 | goto done; | |
3332 | ||
cdfbabfb | 3333 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3334 | if (err < 0) { |
3335 | sock_release(*newsock); | |
fa8705b0 | 3336 | *newsock = NULL; |
ac5a488e SS |
3337 | goto done; |
3338 | } | |
3339 | ||
3340 | (*newsock)->ops = sock->ops; | |
1b08534e | 3341 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3342 | |
3343 | done: | |
3344 | return err; | |
3345 | } | |
c6d409cf | 3346 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e SS |
3347 | |
3348 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, | |
4768fbcb | 3349 | int flags) |
ac5a488e SS |
3350 | { |
3351 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3352 | } | |
c6d409cf | 3353 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e SS |
3354 | |
3355 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr, | |
3356 | int *addrlen) | |
3357 | { | |
3358 | return sock->ops->getname(sock, addr, addrlen, 0); | |
3359 | } | |
c6d409cf | 3360 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e SS |
3361 | |
3362 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr, | |
3363 | int *addrlen) | |
3364 | { | |
3365 | return sock->ops->getname(sock, addr, addrlen, 1); | |
3366 | } | |
c6d409cf | 3367 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e SS |
3368 | |
3369 | int kernel_getsockopt(struct socket *sock, int level, int optname, | |
3370 | char *optval, int *optlen) | |
3371 | { | |
3372 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3373 | char __user *uoptval; |
3374 | int __user *uoptlen; | |
ac5a488e SS |
3375 | int err; |
3376 | ||
fb8621bb NK |
3377 | uoptval = (char __user __force *) optval; |
3378 | uoptlen = (int __user __force *) optlen; | |
3379 | ||
ac5a488e SS |
3380 | set_fs(KERNEL_DS); |
3381 | if (level == SOL_SOCKET) | |
fb8621bb | 3382 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3383 | else |
fb8621bb NK |
3384 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3385 | uoptlen); | |
ac5a488e SS |
3386 | set_fs(oldfs); |
3387 | return err; | |
3388 | } | |
c6d409cf | 3389 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e SS |
3390 | |
3391 | int kernel_setsockopt(struct socket *sock, int level, int optname, | |
b7058842 | 3392 | char *optval, unsigned int optlen) |
ac5a488e SS |
3393 | { |
3394 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3395 | char __user *uoptval; |
ac5a488e SS |
3396 | int err; |
3397 | ||
fb8621bb NK |
3398 | uoptval = (char __user __force *) optval; |
3399 | ||
ac5a488e SS |
3400 | set_fs(KERNEL_DS); |
3401 | if (level == SOL_SOCKET) | |
fb8621bb | 3402 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3403 | else |
fb8621bb | 3404 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3405 | optlen); |
3406 | set_fs(oldfs); | |
3407 | return err; | |
3408 | } | |
c6d409cf | 3409 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e SS |
3410 | |
3411 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, | |
3412 | size_t size, int flags) | |
3413 | { | |
3414 | if (sock->ops->sendpage) | |
3415 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3416 | ||
3417 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3418 | } | |
c6d409cf | 3419 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3420 | |
306b13eb TH |
3421 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3422 | size_t size, int flags) | |
3423 | { | |
3424 | struct socket *sock = sk->sk_socket; | |
3425 | ||
3426 | if (sock->ops->sendpage_locked) | |
3427 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3428 | flags); | |
3429 | ||
3430 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3431 | } | |
3432 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3433 | ||
ac5a488e SS |
3434 | int kernel_sock_ioctl(struct socket *sock, int cmd, unsigned long arg) |
3435 | { | |
3436 | mm_segment_t oldfs = get_fs(); | |
3437 | int err; | |
3438 | ||
3439 | set_fs(KERNEL_DS); | |
3440 | err = sock->ops->ioctl(sock, cmd, arg); | |
3441 | set_fs(oldfs); | |
3442 | ||
3443 | return err; | |
3444 | } | |
c6d409cf | 3445 | EXPORT_SYMBOL(kernel_sock_ioctl); |
ac5a488e | 3446 | |
91cf45f0 TM |
3447 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3448 | { | |
3449 | return sock->ops->shutdown(sock, how); | |
3450 | } | |
91cf45f0 | 3451 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 P |
3452 | |
3453 | /* This routine returns the IP overhead imposed by a socket i.e. | |
3454 | * the length of the underlying IP header, depending on whether | |
3455 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
57240d00 | 3456 | * on at the socket. Assumes that the caller has a lock on the socket. |
113c3075 P |
3457 | */ |
3458 | u32 kernel_sock_ip_overhead(struct sock *sk) | |
3459 | { | |
3460 | struct inet_sock *inet; | |
3461 | struct ip_options_rcu *opt; | |
3462 | u32 overhead = 0; | |
113c3075 P |
3463 | #if IS_ENABLED(CONFIG_IPV6) |
3464 | struct ipv6_pinfo *np; | |
3465 | struct ipv6_txoptions *optv6 = NULL; | |
3466 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3467 | ||
3468 | if (!sk) | |
3469 | return overhead; | |
3470 | ||
113c3075 P |
3471 | switch (sk->sk_family) { |
3472 | case AF_INET: | |
3473 | inet = inet_sk(sk); | |
3474 | overhead += sizeof(struct iphdr); | |
3475 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3476 | sock_owned_by_user(sk)); |
113c3075 P |
3477 | if (opt) |
3478 | overhead += opt->opt.optlen; | |
3479 | return overhead; | |
3480 | #if IS_ENABLED(CONFIG_IPV6) | |
3481 | case AF_INET6: | |
3482 | np = inet6_sk(sk); | |
3483 | overhead += sizeof(struct ipv6hdr); | |
3484 | if (np) | |
3485 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3486 | sock_owned_by_user(sk)); |
113c3075 P |
3487 | if (optv6) |
3488 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3489 | return overhead; | |
3490 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3491 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3492 | return overhead; | |
3493 | } | |
3494 | } | |
3495 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |