]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * NET An implementation of the SOCKET network access protocol. | |
3 | * | |
4 | * Version: @(#)socket.c 1.1.93 18/02/95 | |
5 | * | |
6 | * Authors: Orest Zborowski, <obz@Kodak.COM> | |
02c30a84 | 7 | * Ross Biro |
1da177e4 LT |
8 | * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> |
9 | * | |
10 | * Fixes: | |
11 | * Anonymous : NOTSOCK/BADF cleanup. Error fix in | |
12 | * shutdown() | |
13 | * Alan Cox : verify_area() fixes | |
14 | * Alan Cox : Removed DDI | |
15 | * Jonathan Kamens : SOCK_DGRAM reconnect bug | |
16 | * Alan Cox : Moved a load of checks to the very | |
17 | * top level. | |
18 | * Alan Cox : Move address structures to/from user | |
19 | * mode above the protocol layers. | |
20 | * Rob Janssen : Allow 0 length sends. | |
21 | * Alan Cox : Asynchronous I/O support (cribbed from the | |
22 | * tty drivers). | |
23 | * Niibe Yutaka : Asynchronous I/O for writes (4.4BSD style) | |
24 | * Jeff Uphoff : Made max number of sockets command-line | |
25 | * configurable. | |
26 | * Matti Aarnio : Made the number of sockets dynamic, | |
27 | * to be allocated when needed, and mr. | |
28 | * Uphoff's max is used as max to be | |
29 | * allowed to allocate. | |
30 | * Linus : Argh. removed all the socket allocation | |
31 | * altogether: it's in the inode now. | |
32 | * Alan Cox : Made sock_alloc()/sock_release() public | |
33 | * for NetROM and future kernel nfsd type | |
34 | * stuff. | |
35 | * Alan Cox : sendmsg/recvmsg basics. | |
36 | * Tom Dyas : Export net symbols. | |
37 | * Marcin Dalecki : Fixed problems with CONFIG_NET="n". | |
38 | * Alan Cox : Added thread locking to sys_* calls | |
39 | * for sockets. May have errors at the | |
40 | * moment. | |
41 | * Kevin Buhr : Fixed the dumb errors in the above. | |
42 | * Andi Kleen : Some small cleanups, optimizations, | |
43 | * and fixed a copy_from_user() bug. | |
44 | * Tigran Aivazian : sys_send(args) calls sys_sendto(args, NULL, 0) | |
89bddce5 | 45 | * Tigran Aivazian : Made listen(2) backlog sanity checks |
1da177e4 LT |
46 | * protocol-independent |
47 | * | |
48 | * | |
49 | * This program is free software; you can redistribute it and/or | |
50 | * modify it under the terms of the GNU General Public License | |
51 | * as published by the Free Software Foundation; either version | |
52 | * 2 of the License, or (at your option) any later version. | |
53 | * | |
54 | * | |
55 | * This module is effectively the top level interface to the BSD socket | |
89bddce5 | 56 | * paradigm. |
1da177e4 LT |
57 | * |
58 | * Based upon Swansea University Computer Society NET3.039 | |
59 | */ | |
60 | ||
1da177e4 | 61 | #include <linux/mm.h> |
1da177e4 LT |
62 | #include <linux/socket.h> |
63 | #include <linux/file.h> | |
64 | #include <linux/net.h> | |
65 | #include <linux/interrupt.h> | |
aaca0bdc | 66 | #include <linux/thread_info.h> |
55737fda | 67 | #include <linux/rcupdate.h> |
1da177e4 LT |
68 | #include <linux/netdevice.h> |
69 | #include <linux/proc_fs.h> | |
70 | #include <linux/seq_file.h> | |
4a3e2f71 | 71 | #include <linux/mutex.h> |
1da177e4 | 72 | #include <linux/if_bridge.h> |
20380731 ACM |
73 | #include <linux/if_frad.h> |
74 | #include <linux/if_vlan.h> | |
408eccce | 75 | #include <linux/ptp_classify.h> |
1da177e4 LT |
76 | #include <linux/init.h> |
77 | #include <linux/poll.h> | |
78 | #include <linux/cache.h> | |
79 | #include <linux/module.h> | |
80 | #include <linux/highmem.h> | |
1da177e4 LT |
81 | #include <linux/mount.h> |
82 | #include <linux/security.h> | |
83 | #include <linux/syscalls.h> | |
84 | #include <linux/compat.h> | |
85 | #include <linux/kmod.h> | |
3ec3b2fb | 86 | #include <linux/audit.h> |
d86b5e0e | 87 | #include <linux/wireless.h> |
1b8d7ae4 | 88 | #include <linux/nsproxy.h> |
1fd7317d | 89 | #include <linux/magic.h> |
5a0e3ad6 | 90 | #include <linux/slab.h> |
600e1779 | 91 | #include <linux/xattr.h> |
1da177e4 | 92 | |
7c0f6ba6 | 93 | #include <linux/uaccess.h> |
1da177e4 LT |
94 | #include <asm/unistd.h> |
95 | ||
96 | #include <net/compat.h> | |
87de87d5 | 97 | #include <net/wext.h> |
f8451725 | 98 | #include <net/cls_cgroup.h> |
1da177e4 LT |
99 | |
100 | #include <net/sock.h> | |
101 | #include <linux/netfilter.h> | |
102 | ||
6b96018b AB |
103 | #include <linux/if_tun.h> |
104 | #include <linux/ipv6_route.h> | |
105 | #include <linux/route.h> | |
6b96018b AB |
106 | #include <linux/sockios.h> |
107 | #include <linux/atalk.h> | |
076bb0c8 | 108 | #include <net/busy_poll.h> |
f24b9be5 | 109 | #include <linux/errqueue.h> |
06021292 | 110 | |
e0d1095a | 111 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
112 | unsigned int sysctl_net_busy_read __read_mostly; |
113 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 114 | #endif |
6b96018b | 115 | |
8ae5e030 AV |
116 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
117 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 118 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
119 | |
120 | static int sock_close(struct inode *inode, struct file *file); | |
ade994f4 | 121 | static __poll_t sock_poll(struct file *file, |
1da177e4 | 122 | struct poll_table_struct *wait); |
89bddce5 | 123 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
124 | #ifdef CONFIG_COMPAT |
125 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 126 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 127 | #endif |
1da177e4 | 128 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
129 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
130 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 131 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 132 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 133 | unsigned int flags); |
1da177e4 | 134 | |
1da177e4 LT |
135 | /* |
136 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
137 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
138 | */ | |
139 | ||
da7071d7 | 140 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
141 | .owner = THIS_MODULE, |
142 | .llseek = no_llseek, | |
8ae5e030 AV |
143 | .read_iter = sock_read_iter, |
144 | .write_iter = sock_write_iter, | |
1da177e4 LT |
145 | .poll = sock_poll, |
146 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
147 | #ifdef CONFIG_COMPAT |
148 | .compat_ioctl = compat_sock_ioctl, | |
149 | #endif | |
1da177e4 | 150 | .mmap = sock_mmap, |
1da177e4 LT |
151 | .release = sock_close, |
152 | .fasync = sock_fasync, | |
5274f052 JA |
153 | .sendpage = sock_sendpage, |
154 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 155 | .splice_read = sock_splice_read, |
1da177e4 LT |
156 | }; |
157 | ||
158 | /* | |
159 | * The protocol list. Each protocol is registered in here. | |
160 | */ | |
161 | ||
1da177e4 | 162 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 163 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 164 | |
1da177e4 | 165 | /* |
89bddce5 SH |
166 | * Support routines. |
167 | * Move socket addresses back and forth across the kernel/user | |
168 | * divide and look after the messy bits. | |
1da177e4 LT |
169 | */ |
170 | ||
1da177e4 LT |
171 | /** |
172 | * move_addr_to_kernel - copy a socket address into kernel space | |
173 | * @uaddr: Address in user space | |
174 | * @kaddr: Address in kernel space | |
175 | * @ulen: Length in user space | |
176 | * | |
177 | * The address is copied into kernel space. If the provided address is | |
178 | * too long an error code of -EINVAL is returned. If the copy gives | |
179 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
180 | */ | |
181 | ||
43db362d | 182 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 183 | { |
230b1839 | 184 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 185 | return -EINVAL; |
89bddce5 | 186 | if (ulen == 0) |
1da177e4 | 187 | return 0; |
89bddce5 | 188 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 189 | return -EFAULT; |
3ec3b2fb | 190 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
191 | } |
192 | ||
193 | /** | |
194 | * move_addr_to_user - copy an address to user space | |
195 | * @kaddr: kernel space address | |
196 | * @klen: length of address in kernel | |
197 | * @uaddr: user space address | |
198 | * @ulen: pointer to user length field | |
199 | * | |
200 | * The value pointed to by ulen on entry is the buffer length available. | |
201 | * This is overwritten with the buffer space used. -EINVAL is returned | |
202 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
203 | * is returned if either the buffer or the length field are not | |
204 | * accessible. | |
205 | * After copying the data up to the limit the user specifies, the true | |
206 | * length of the data is written over the length limit the user | |
207 | * specified. Zero is returned for a success. | |
208 | */ | |
89bddce5 | 209 | |
43db362d | 210 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 211 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
212 | { |
213 | int err; | |
214 | int len; | |
215 | ||
68c6beb3 | 216 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
217 | err = get_user(len, ulen); |
218 | if (err) | |
1da177e4 | 219 | return err; |
89bddce5 SH |
220 | if (len > klen) |
221 | len = klen; | |
68c6beb3 | 222 | if (len < 0) |
1da177e4 | 223 | return -EINVAL; |
89bddce5 | 224 | if (len) { |
d6fe3945 SG |
225 | if (audit_sockaddr(klen, kaddr)) |
226 | return -ENOMEM; | |
89bddce5 | 227 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
228 | return -EFAULT; |
229 | } | |
230 | /* | |
89bddce5 SH |
231 | * "fromlen shall refer to the value before truncation.." |
232 | * 1003.1g | |
1da177e4 LT |
233 | */ |
234 | return __put_user(klen, ulen); | |
235 | } | |
236 | ||
e18b890b | 237 | static struct kmem_cache *sock_inode_cachep __read_mostly; |
1da177e4 LT |
238 | |
239 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
240 | { | |
241 | struct socket_alloc *ei; | |
eaefd110 | 242 | struct socket_wq *wq; |
89bddce5 | 243 | |
e94b1766 | 244 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
245 | if (!ei) |
246 | return NULL; | |
eaefd110 ED |
247 | wq = kmalloc(sizeof(*wq), GFP_KERNEL); |
248 | if (!wq) { | |
43815482 ED |
249 | kmem_cache_free(sock_inode_cachep, ei); |
250 | return NULL; | |
251 | } | |
eaefd110 ED |
252 | init_waitqueue_head(&wq->wait); |
253 | wq->fasync_list = NULL; | |
574aab1e | 254 | wq->flags = 0; |
eaefd110 | 255 | RCU_INIT_POINTER(ei->socket.wq, wq); |
89bddce5 | 256 | |
1da177e4 LT |
257 | ei->socket.state = SS_UNCONNECTED; |
258 | ei->socket.flags = 0; | |
259 | ei->socket.ops = NULL; | |
260 | ei->socket.sk = NULL; | |
261 | ei->socket.file = NULL; | |
1da177e4 LT |
262 | |
263 | return &ei->vfs_inode; | |
264 | } | |
265 | ||
266 | static void sock_destroy_inode(struct inode *inode) | |
267 | { | |
43815482 | 268 | struct socket_alloc *ei; |
eaefd110 | 269 | struct socket_wq *wq; |
43815482 ED |
270 | |
271 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
eaefd110 | 272 | wq = rcu_dereference_protected(ei->socket.wq, 1); |
61845220 | 273 | kfree_rcu(wq, rcu); |
43815482 | 274 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
275 | } |
276 | ||
51cc5068 | 277 | static void init_once(void *foo) |
1da177e4 | 278 | { |
89bddce5 | 279 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 280 | |
a35afb83 | 281 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 282 | } |
89bddce5 | 283 | |
1e911632 | 284 | static void init_inodecache(void) |
1da177e4 LT |
285 | { |
286 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
287 | sizeof(struct socket_alloc), |
288 | 0, | |
289 | (SLAB_HWCACHE_ALIGN | | |
290 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 291 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 292 | init_once); |
1e911632 | 293 | BUG_ON(sock_inode_cachep == NULL); |
1da177e4 LT |
294 | } |
295 | ||
b87221de | 296 | static const struct super_operations sockfs_ops = { |
c6d409cf ED |
297 | .alloc_inode = sock_alloc_inode, |
298 | .destroy_inode = sock_destroy_inode, | |
299 | .statfs = simple_statfs, | |
1da177e4 LT |
300 | }; |
301 | ||
c23fbb6b ED |
302 | /* |
303 | * sockfs_dname() is called from d_path(). | |
304 | */ | |
305 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
306 | { | |
307 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 308 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
309 | } |
310 | ||
3ba13d17 | 311 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 312 | .d_dname = sockfs_dname, |
1da177e4 LT |
313 | }; |
314 | ||
bba0bd31 AG |
315 | static int sockfs_xattr_get(const struct xattr_handler *handler, |
316 | struct dentry *dentry, struct inode *inode, | |
317 | const char *suffix, void *value, size_t size) | |
318 | { | |
319 | if (value) { | |
320 | if (dentry->d_name.len + 1 > size) | |
321 | return -ERANGE; | |
322 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
323 | } | |
324 | return dentry->d_name.len + 1; | |
325 | } | |
326 | ||
327 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" | |
328 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
329 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
330 | ||
331 | static const struct xattr_handler sockfs_xattr_handler = { | |
332 | .name = XATTR_NAME_SOCKPROTONAME, | |
333 | .get = sockfs_xattr_get, | |
334 | }; | |
335 | ||
4a590153 AG |
336 | static int sockfs_security_xattr_set(const struct xattr_handler *handler, |
337 | struct dentry *dentry, struct inode *inode, | |
338 | const char *suffix, const void *value, | |
339 | size_t size, int flags) | |
340 | { | |
341 | /* Handled by LSM. */ | |
342 | return -EAGAIN; | |
343 | } | |
344 | ||
345 | static const struct xattr_handler sockfs_security_xattr_handler = { | |
346 | .prefix = XATTR_SECURITY_PREFIX, | |
347 | .set = sockfs_security_xattr_set, | |
348 | }; | |
349 | ||
bba0bd31 AG |
350 | static const struct xattr_handler *sockfs_xattr_handlers[] = { |
351 | &sockfs_xattr_handler, | |
4a590153 | 352 | &sockfs_security_xattr_handler, |
bba0bd31 AG |
353 | NULL |
354 | }; | |
355 | ||
c74a1cbb AV |
356 | static struct dentry *sockfs_mount(struct file_system_type *fs_type, |
357 | int flags, const char *dev_name, void *data) | |
358 | { | |
bba0bd31 AG |
359 | return mount_pseudo_xattr(fs_type, "socket:", &sockfs_ops, |
360 | sockfs_xattr_handlers, | |
361 | &sockfs_dentry_operations, SOCKFS_MAGIC); | |
c74a1cbb AV |
362 | } |
363 | ||
364 | static struct vfsmount *sock_mnt __read_mostly; | |
365 | ||
366 | static struct file_system_type sock_fs_type = { | |
367 | .name = "sockfs", | |
368 | .mount = sockfs_mount, | |
369 | .kill_sb = kill_anon_super, | |
370 | }; | |
371 | ||
1da177e4 LT |
372 | /* |
373 | * Obtains the first available file descriptor and sets it up for use. | |
374 | * | |
39d8c1b6 DM |
375 | * These functions create file structures and maps them to fd space |
376 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
377 | * and file struct implicitly stored in sock->file. |
378 | * Note that another thread may close file descriptor before we return | |
379 | * from this function. We use the fact that now we do not refer | |
380 | * to socket after mapping. If one day we will need it, this | |
381 | * function will increment ref. count on file by 1. | |
382 | * | |
383 | * In any case returned fd MAY BE not valid! | |
384 | * This race condition is unavoidable | |
385 | * with shared fd spaces, we cannot solve it inside kernel, | |
386 | * but we take care of internal coherence yet. | |
387 | */ | |
388 | ||
aab174f0 | 389 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 390 | { |
7cbe66b6 | 391 | struct qstr name = { .name = "" }; |
2c48b9c4 | 392 | struct path path; |
7cbe66b6 | 393 | struct file *file; |
1da177e4 | 394 | |
600e1779 MY |
395 | if (dname) { |
396 | name.name = dname; | |
397 | name.len = strlen(name.name); | |
398 | } else if (sock->sk) { | |
399 | name.name = sock->sk->sk_prot_creator->name; | |
400 | name.len = strlen(name.name); | |
401 | } | |
4b936885 | 402 | path.dentry = d_alloc_pseudo(sock_mnt->mnt_sb, &name); |
8e1611e2 AV |
403 | if (unlikely(!path.dentry)) { |
404 | sock_release(sock); | |
28407630 | 405 | return ERR_PTR(-ENOMEM); |
8e1611e2 | 406 | } |
2c48b9c4 | 407 | path.mnt = mntget(sock_mnt); |
39d8c1b6 | 408 | |
2c48b9c4 | 409 | d_instantiate(path.dentry, SOCK_INODE(sock)); |
39d8c1b6 | 410 | |
2c48b9c4 | 411 | file = alloc_file(&path, FMODE_READ | FMODE_WRITE, |
ce8d2cdf | 412 | &socket_file_ops); |
b5ffe634 | 413 | if (IS_ERR(file)) { |
8e1611e2 | 414 | /* drop dentry, keep inode for a bit */ |
c5ef6035 | 415 | ihold(d_inode(path.dentry)); |
2c48b9c4 | 416 | path_put(&path); |
8e1611e2 AV |
417 | /* ... and now kill it properly */ |
418 | sock_release(sock); | |
39b65252 | 419 | return file; |
cc3808f8 AV |
420 | } |
421 | ||
422 | sock->file = file; | |
77d27200 | 423 | file->f_flags = O_RDWR | (flags & O_NONBLOCK); |
39d8c1b6 | 424 | file->private_data = sock; |
28407630 | 425 | return file; |
39d8c1b6 | 426 | } |
56b31d1c | 427 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 428 | |
56b31d1c | 429 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
430 | { |
431 | struct file *newfile; | |
28407630 | 432 | int fd = get_unused_fd_flags(flags); |
ce4bb04c AV |
433 | if (unlikely(fd < 0)) { |
434 | sock_release(sock); | |
28407630 | 435 | return fd; |
ce4bb04c | 436 | } |
39d8c1b6 | 437 | |
aab174f0 | 438 | newfile = sock_alloc_file(sock, flags, NULL); |
28407630 | 439 | if (likely(!IS_ERR(newfile))) { |
39d8c1b6 | 440 | fd_install(fd, newfile); |
28407630 AV |
441 | return fd; |
442 | } | |
7cbe66b6 | 443 | |
28407630 AV |
444 | put_unused_fd(fd); |
445 | return PTR_ERR(newfile); | |
1da177e4 LT |
446 | } |
447 | ||
406a3c63 | 448 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 449 | { |
6cb153ca BL |
450 | if (file->f_op == &socket_file_ops) |
451 | return file->private_data; /* set in sock_map_fd */ | |
452 | ||
23bb80d2 ED |
453 | *err = -ENOTSOCK; |
454 | return NULL; | |
6cb153ca | 455 | } |
406a3c63 | 456 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 457 | |
1da177e4 | 458 | /** |
c6d409cf | 459 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
460 | * @fd: file handle |
461 | * @err: pointer to an error code return | |
462 | * | |
463 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 464 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
465 | * with a negative errno code and NULL is returned. The function checks |
466 | * for both invalid handles and passing a handle which is not a socket. | |
467 | * | |
468 | * On a success the socket object pointer is returned. | |
469 | */ | |
470 | ||
471 | struct socket *sockfd_lookup(int fd, int *err) | |
472 | { | |
473 | struct file *file; | |
1da177e4 LT |
474 | struct socket *sock; |
475 | ||
89bddce5 SH |
476 | file = fget(fd); |
477 | if (!file) { | |
1da177e4 LT |
478 | *err = -EBADF; |
479 | return NULL; | |
480 | } | |
89bddce5 | 481 | |
6cb153ca BL |
482 | sock = sock_from_file(file, err); |
483 | if (!sock) | |
1da177e4 | 484 | fput(file); |
6cb153ca BL |
485 | return sock; |
486 | } | |
c6d409cf | 487 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 488 | |
6cb153ca BL |
489 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
490 | { | |
00e188ef | 491 | struct fd f = fdget(fd); |
6cb153ca BL |
492 | struct socket *sock; |
493 | ||
3672558c | 494 | *err = -EBADF; |
00e188ef AV |
495 | if (f.file) { |
496 | sock = sock_from_file(f.file, err); | |
497 | if (likely(sock)) { | |
498 | *fput_needed = f.flags; | |
6cb153ca | 499 | return sock; |
00e188ef AV |
500 | } |
501 | fdput(f); | |
1da177e4 | 502 | } |
6cb153ca | 503 | return NULL; |
1da177e4 LT |
504 | } |
505 | ||
600e1779 MY |
506 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
507 | size_t size) | |
508 | { | |
509 | ssize_t len; | |
510 | ssize_t used = 0; | |
511 | ||
c5ef6035 | 512 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
513 | if (len < 0) |
514 | return len; | |
515 | used += len; | |
516 | if (buffer) { | |
517 | if (size < used) | |
518 | return -ERANGE; | |
519 | buffer += len; | |
520 | } | |
521 | ||
522 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
523 | used += len; | |
524 | if (buffer) { | |
525 | if (size < used) | |
526 | return -ERANGE; | |
527 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
528 | buffer += len; | |
529 | } | |
530 | ||
531 | return used; | |
532 | } | |
533 | ||
dc647ec8 | 534 | static int sockfs_setattr(struct dentry *dentry, struct iattr *iattr) |
86741ec2 LC |
535 | { |
536 | int err = simple_setattr(dentry, iattr); | |
537 | ||
e1a3a60a | 538 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
539 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
540 | ||
541 | sock->sk->sk_uid = iattr->ia_uid; | |
542 | } | |
543 | ||
544 | return err; | |
545 | } | |
546 | ||
600e1779 | 547 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 548 | .listxattr = sockfs_listxattr, |
86741ec2 | 549 | .setattr = sockfs_setattr, |
600e1779 MY |
550 | }; |
551 | ||
1da177e4 LT |
552 | /** |
553 | * sock_alloc - allocate a socket | |
89bddce5 | 554 | * |
1da177e4 LT |
555 | * Allocate a new inode and socket object. The two are bound together |
556 | * and initialised. The socket is then returned. If we are out of inodes | |
557 | * NULL is returned. | |
558 | */ | |
559 | ||
f4a00aac | 560 | struct socket *sock_alloc(void) |
1da177e4 | 561 | { |
89bddce5 SH |
562 | struct inode *inode; |
563 | struct socket *sock; | |
1da177e4 | 564 | |
a209dfc7 | 565 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
566 | if (!inode) |
567 | return NULL; | |
568 | ||
569 | sock = SOCKET_I(inode); | |
570 | ||
85fe4025 | 571 | inode->i_ino = get_next_ino(); |
89bddce5 | 572 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
573 | inode->i_uid = current_fsuid(); |
574 | inode->i_gid = current_fsgid(); | |
600e1779 | 575 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 576 | |
1da177e4 LT |
577 | return sock; |
578 | } | |
f4a00aac | 579 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 580 | |
1da177e4 LT |
581 | /** |
582 | * sock_release - close a socket | |
583 | * @sock: socket to close | |
584 | * | |
585 | * The socket is released from the protocol stack if it has a release | |
586 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 587 | * an inode not a file. |
1da177e4 | 588 | */ |
89bddce5 | 589 | |
1da177e4 LT |
590 | void sock_release(struct socket *sock) |
591 | { | |
592 | if (sock->ops) { | |
593 | struct module *owner = sock->ops->owner; | |
594 | ||
595 | sock->ops->release(sock); | |
596 | sock->ops = NULL; | |
597 | module_put(owner); | |
598 | } | |
599 | ||
eaefd110 | 600 | if (rcu_dereference_protected(sock->wq, 1)->fasync_list) |
3410f22e | 601 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 602 | |
1da177e4 LT |
603 | if (!sock->file) { |
604 | iput(SOCK_INODE(sock)); | |
605 | return; | |
606 | } | |
89bddce5 | 607 | sock->file = NULL; |
1da177e4 | 608 | } |
c6d409cf | 609 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 610 | |
c14ac945 | 611 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 612 | { |
140c55d4 ED |
613 | u8 flags = *tx_flags; |
614 | ||
c14ac945 | 615 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
616 | flags |= SKBTX_HW_TSTAMP; |
617 | ||
c14ac945 | 618 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
619 | flags |= SKBTX_SW_TSTAMP; |
620 | ||
c14ac945 | 621 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
622 | flags |= SKBTX_SCHED_TSTAMP; |
623 | ||
140c55d4 | 624 | *tx_flags = flags; |
20d49473 | 625 | } |
67cc0d40 | 626 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 627 | |
d8725c86 | 628 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 629 | { |
01e97e65 | 630 | int ret = sock->ops->sendmsg(sock, msg, msg_data_left(msg)); |
d8725c86 AV |
631 | BUG_ON(ret == -EIOCBQUEUED); |
632 | return ret; | |
1da177e4 LT |
633 | } |
634 | ||
d8725c86 | 635 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 636 | { |
d8725c86 | 637 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 638 | msg_data_left(msg)); |
228e548e | 639 | |
d8725c86 | 640 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 641 | } |
c6d409cf | 642 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 LT |
643 | |
644 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, | |
645 | struct kvec *vec, size_t num, size_t size) | |
646 | { | |
6aa24814 | 647 | iov_iter_kvec(&msg->msg_iter, WRITE | ITER_KVEC, vec, num, size); |
d8725c86 | 648 | return sock_sendmsg(sock, msg); |
1da177e4 | 649 | } |
c6d409cf | 650 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 651 | |
306b13eb TH |
652 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
653 | struct kvec *vec, size_t num, size_t size) | |
654 | { | |
655 | struct socket *sock = sk->sk_socket; | |
656 | ||
657 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 658 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb TH |
659 | |
660 | iov_iter_kvec(&msg->msg_iter, WRITE | ITER_KVEC, vec, num, size); | |
661 | ||
662 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
663 | } | |
664 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
665 | ||
8605330a SHY |
666 | static bool skb_is_err_queue(const struct sk_buff *skb) |
667 | { | |
668 | /* pkt_type of skbs enqueued on the error queue are set to | |
669 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
670 | * in recvmsg, since skbs received on a local socket will never | |
671 | * have a pkt_type of PACKET_OUTGOING. | |
672 | */ | |
673 | return skb->pkt_type == PACKET_OUTGOING; | |
674 | } | |
675 | ||
b50a5c70 ML |
676 | /* On transmit, software and hardware timestamps are returned independently. |
677 | * As the two skb clones share the hardware timestamp, which may be updated | |
678 | * before the software timestamp is received, a hardware TX timestamp may be | |
679 | * returned only if there is no software TX timestamp. Ignore false software | |
680 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
681 | * option SO_TIMESTAMP(NS) is enabled on the socket, even when the skb has a | |
682 | * hardware timestamp. | |
683 | */ | |
684 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
685 | { | |
686 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
687 | } | |
688 | ||
aad9c8c4 ML |
689 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb) |
690 | { | |
691 | struct scm_ts_pktinfo ts_pktinfo; | |
692 | struct net_device *orig_dev; | |
693 | ||
694 | if (!skb_mac_header_was_set(skb)) | |
695 | return; | |
696 | ||
697 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
698 | ||
699 | rcu_read_lock(); | |
700 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
701 | if (orig_dev) | |
702 | ts_pktinfo.if_index = orig_dev->ifindex; | |
703 | rcu_read_unlock(); | |
704 | ||
705 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
706 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
707 | sizeof(ts_pktinfo), &ts_pktinfo); | |
708 | } | |
709 | ||
92f37fd2 ED |
710 | /* |
711 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
712 | */ | |
713 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
714 | struct sk_buff *skb) | |
715 | { | |
20d49473 | 716 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
f24b9be5 | 717 | struct scm_timestamping tss; |
b50a5c70 | 718 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
719 | struct skb_shared_hwtstamps *shhwtstamps = |
720 | skb_hwtstamps(skb); | |
721 | ||
722 | /* Race occurred between timestamp enabling and packet | |
723 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 724 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 725 | __net_timestamp(skb); |
b50a5c70 ML |
726 | false_tstamp = 1; |
727 | } | |
20d49473 PO |
728 | |
729 | if (need_software_tstamp) { | |
730 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
731 | struct timeval tv; | |
732 | skb_get_timestamp(skb, &tv); | |
733 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMP, | |
734 | sizeof(tv), &tv); | |
735 | } else { | |
f24b9be5 WB |
736 | struct timespec ts; |
737 | skb_get_timestampns(skb, &ts); | |
20d49473 | 738 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPNS, |
f24b9be5 | 739 | sizeof(ts), &ts); |
20d49473 PO |
740 | } |
741 | } | |
742 | ||
f24b9be5 | 743 | memset(&tss, 0, sizeof(tss)); |
c199105d | 744 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
f24b9be5 | 745 | ktime_to_timespec_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 746 | empty = 0; |
4d276eb6 | 747 | if (shhwtstamps && |
b9f40e21 | 748 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
b50a5c70 | 749 | !skb_is_swtx_tstamp(skb, false_tstamp) && |
aad9c8c4 | 750 | ktime_to_timespec_cond(shhwtstamps->hwtstamp, tss.ts + 2)) { |
4d276eb6 | 751 | empty = 0; |
aad9c8c4 ML |
752 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && |
753 | !skb_is_err_queue(skb)) | |
754 | put_ts_pktinfo(msg, skb); | |
755 | } | |
1c885808 | 756 | if (!empty) { |
20d49473 | 757 | put_cmsg(msg, SOL_SOCKET, |
f24b9be5 | 758 | SCM_TIMESTAMPING, sizeof(tss), &tss); |
1c885808 | 759 | |
8605330a | 760 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 761 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
762 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
763 | skb->len, skb->data); | |
764 | } | |
92f37fd2 | 765 | } |
7c81fd8b ACM |
766 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
767 | ||
6e3e939f JB |
768 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
769 | struct sk_buff *skb) | |
770 | { | |
771 | int ack; | |
772 | ||
773 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
774 | return; | |
775 | if (!skb->wifi_acked_valid) | |
776 | return; | |
777 | ||
778 | ack = skb->wifi_acked; | |
779 | ||
780 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
781 | } | |
782 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
783 | ||
11165f14 | 784 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
785 | struct sk_buff *skb) | |
3b885787 | 786 | { |
744d5a3e | 787 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 788 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 789 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
790 | } |
791 | ||
767dd033 | 792 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
793 | struct sk_buff *skb) |
794 | { | |
795 | sock_recv_timestamp(msg, sk, skb); | |
796 | sock_recv_drops(msg, sk, skb); | |
797 | } | |
767dd033 | 798 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 799 | |
1b784140 | 800 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 801 | int flags) |
1da177e4 | 802 | { |
2da62906 | 803 | return sock->ops->recvmsg(sock, msg, msg_data_left(msg), flags); |
1da177e4 LT |
804 | } |
805 | ||
2da62906 | 806 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 807 | { |
2da62906 | 808 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 809 | |
2da62906 | 810 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 811 | } |
c6d409cf | 812 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 813 | |
c1249c0a ML |
814 | /** |
815 | * kernel_recvmsg - Receive a message from a socket (kernel space) | |
816 | * @sock: The socket to receive the message from | |
817 | * @msg: Received message | |
818 | * @vec: Input s/g array for message data | |
819 | * @num: Size of input s/g array | |
820 | * @size: Number of bytes to read | |
821 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
822 | * | |
823 | * On return the msg structure contains the scatter/gather array passed in the | |
824 | * vec argument. The array is modified so that it consists of the unfilled | |
825 | * portion of the original array. | |
826 | * | |
827 | * The returned value is the total number of bytes received, or an error. | |
828 | */ | |
89bddce5 SH |
829 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
830 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
831 | { |
832 | mm_segment_t oldfs = get_fs(); | |
833 | int result; | |
834 | ||
6aa24814 | 835 | iov_iter_kvec(&msg->msg_iter, READ | ITER_KVEC, vec, num, size); |
1da177e4 | 836 | set_fs(KERNEL_DS); |
2da62906 | 837 | result = sock_recvmsg(sock, msg, flags); |
1da177e4 LT |
838 | set_fs(oldfs); |
839 | return result; | |
840 | } | |
c6d409cf | 841 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 842 | |
ce1d4d3e CH |
843 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
844 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 845 | { |
1da177e4 LT |
846 | struct socket *sock; |
847 | int flags; | |
848 | ||
ce1d4d3e CH |
849 | sock = file->private_data; |
850 | ||
35f9c09f ED |
851 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
852 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
853 | flags |= more; | |
ce1d4d3e | 854 | |
e6949583 | 855 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 856 | } |
1da177e4 | 857 | |
9c55e01c | 858 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 859 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
860 | unsigned int flags) |
861 | { | |
862 | struct socket *sock = file->private_data; | |
863 | ||
997b37da RDC |
864 | if (unlikely(!sock->ops->splice_read)) |
865 | return -EINVAL; | |
866 | ||
9c55e01c JA |
867 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
868 | } | |
869 | ||
8ae5e030 | 870 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 871 | { |
6d652330 AV |
872 | struct file *file = iocb->ki_filp; |
873 | struct socket *sock = file->private_data; | |
0345f931 | 874 | struct msghdr msg = {.msg_iter = *to, |
875 | .msg_iocb = iocb}; | |
8ae5e030 | 876 | ssize_t res; |
ce1d4d3e | 877 | |
8ae5e030 AV |
878 | if (file->f_flags & O_NONBLOCK) |
879 | msg.msg_flags = MSG_DONTWAIT; | |
880 | ||
881 | if (iocb->ki_pos != 0) | |
1da177e4 | 882 | return -ESPIPE; |
027445c3 | 883 | |
66ee59af | 884 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
885 | return 0; |
886 | ||
2da62906 | 887 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
888 | *to = msg.msg_iter; |
889 | return res; | |
1da177e4 LT |
890 | } |
891 | ||
8ae5e030 | 892 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 893 | { |
6d652330 AV |
894 | struct file *file = iocb->ki_filp; |
895 | struct socket *sock = file->private_data; | |
0345f931 | 896 | struct msghdr msg = {.msg_iter = *from, |
897 | .msg_iocb = iocb}; | |
8ae5e030 | 898 | ssize_t res; |
1da177e4 | 899 | |
8ae5e030 | 900 | if (iocb->ki_pos != 0) |
ce1d4d3e | 901 | return -ESPIPE; |
027445c3 | 902 | |
8ae5e030 AV |
903 | if (file->f_flags & O_NONBLOCK) |
904 | msg.msg_flags = MSG_DONTWAIT; | |
905 | ||
6d652330 AV |
906 | if (sock->type == SOCK_SEQPACKET) |
907 | msg.msg_flags |= MSG_EOR; | |
908 | ||
d8725c86 | 909 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
910 | *from = msg.msg_iter; |
911 | return res; | |
1da177e4 LT |
912 | } |
913 | ||
1da177e4 LT |
914 | /* |
915 | * Atomic setting of ioctl hooks to avoid race | |
916 | * with module unload. | |
917 | */ | |
918 | ||
4a3e2f71 | 919 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 920 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 921 | |
881d966b | 922 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 923 | { |
4a3e2f71 | 924 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 925 | br_ioctl_hook = hook; |
4a3e2f71 | 926 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
927 | } |
928 | EXPORT_SYMBOL(brioctl_set); | |
929 | ||
4a3e2f71 | 930 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 931 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 932 | |
881d966b | 933 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 934 | { |
4a3e2f71 | 935 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 936 | vlan_ioctl_hook = hook; |
4a3e2f71 | 937 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
938 | } |
939 | EXPORT_SYMBOL(vlan_ioctl_set); | |
940 | ||
4a3e2f71 | 941 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 942 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 943 | |
89bddce5 | 944 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 945 | { |
4a3e2f71 | 946 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 947 | dlci_ioctl_hook = hook; |
4a3e2f71 | 948 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
949 | } |
950 | EXPORT_SYMBOL(dlci_ioctl_set); | |
951 | ||
6b96018b AB |
952 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
953 | unsigned int cmd, unsigned long arg) | |
954 | { | |
955 | int err; | |
956 | void __user *argp = (void __user *)arg; | |
957 | ||
958 | err = sock->ops->ioctl(sock, cmd, arg); | |
959 | ||
960 | /* | |
961 | * If this ioctl is unknown try to hand it down | |
962 | * to the NIC driver. | |
963 | */ | |
36fd633e AV |
964 | if (err != -ENOIOCTLCMD) |
965 | return err; | |
6b96018b | 966 | |
36fd633e AV |
967 | if (cmd == SIOCGIFCONF) { |
968 | struct ifconf ifc; | |
969 | if (copy_from_user(&ifc, argp, sizeof(struct ifconf))) | |
970 | return -EFAULT; | |
971 | rtnl_lock(); | |
972 | err = dev_ifconf(net, &ifc, sizeof(struct ifreq)); | |
973 | rtnl_unlock(); | |
974 | if (!err && copy_to_user(argp, &ifc, sizeof(struct ifconf))) | |
975 | err = -EFAULT; | |
44c02a2c AV |
976 | } else { |
977 | struct ifreq ifr; | |
978 | bool need_copyout; | |
979 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
980 | return -EFAULT; | |
981 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
982 | if (!err && need_copyout) | |
983 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
984 | return -EFAULT; | |
36fd633e | 985 | } |
6b96018b AB |
986 | return err; |
987 | } | |
988 | ||
1da177e4 LT |
989 | /* |
990 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
991 | * what to do with it - that's up to the protocol still. | |
992 | */ | |
993 | ||
c62cce2c AV |
994 | static struct ns_common *get_net_ns(struct ns_common *ns) |
995 | { | |
996 | return &get_net(container_of(ns, struct net, ns))->ns; | |
997 | } | |
998 | ||
1da177e4 LT |
999 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
1000 | { | |
1001 | struct socket *sock; | |
881d966b | 1002 | struct sock *sk; |
1da177e4 LT |
1003 | void __user *argp = (void __user *)arg; |
1004 | int pid, err; | |
881d966b | 1005 | struct net *net; |
1da177e4 | 1006 | |
b69aee04 | 1007 | sock = file->private_data; |
881d966b | 1008 | sk = sock->sk; |
3b1e0a65 | 1009 | net = sock_net(sk); |
44c02a2c AV |
1010 | if (unlikely(cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))) { |
1011 | struct ifreq ifr; | |
1012 | bool need_copyout; | |
1013 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
1014 | return -EFAULT; | |
1015 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1016 | if (!err && need_copyout) | |
1017 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
1018 | return -EFAULT; | |
1da177e4 | 1019 | } else |
3d23e349 | 1020 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1021 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
b1b0c245 | 1022 | err = wext_handle_ioctl(net, cmd, argp); |
1da177e4 | 1023 | } else |
3d23e349 | 1024 | #endif |
89bddce5 | 1025 | switch (cmd) { |
1da177e4 LT |
1026 | case FIOSETOWN: |
1027 | case SIOCSPGRP: | |
1028 | err = -EFAULT; | |
1029 | if (get_user(pid, (int __user *)argp)) | |
1030 | break; | |
393cc3f5 | 1031 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1032 | break; |
1033 | case FIOGETOWN: | |
1034 | case SIOCGPGRP: | |
609d7fa9 | 1035 | err = put_user(f_getown(sock->file), |
89bddce5 | 1036 | (int __user *)argp); |
1da177e4 LT |
1037 | break; |
1038 | case SIOCGIFBR: | |
1039 | case SIOCSIFBR: | |
1040 | case SIOCBRADDBR: | |
1041 | case SIOCBRDELBR: | |
1042 | err = -ENOPKG; | |
1043 | if (!br_ioctl_hook) | |
1044 | request_module("bridge"); | |
1045 | ||
4a3e2f71 | 1046 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1047 | if (br_ioctl_hook) |
881d966b | 1048 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1049 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1050 | break; |
1051 | case SIOCGIFVLAN: | |
1052 | case SIOCSIFVLAN: | |
1053 | err = -ENOPKG; | |
1054 | if (!vlan_ioctl_hook) | |
1055 | request_module("8021q"); | |
1056 | ||
4a3e2f71 | 1057 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1058 | if (vlan_ioctl_hook) |
881d966b | 1059 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1060 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1061 | break; |
1da177e4 LT |
1062 | case SIOCADDDLCI: |
1063 | case SIOCDELDLCI: | |
1064 | err = -ENOPKG; | |
1065 | if (!dlci_ioctl_hook) | |
1066 | request_module("dlci"); | |
1067 | ||
7512cbf6 PE |
1068 | mutex_lock(&dlci_ioctl_mutex); |
1069 | if (dlci_ioctl_hook) | |
1da177e4 | 1070 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1071 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 | 1072 | break; |
c62cce2c AV |
1073 | case SIOCGSKNS: |
1074 | err = -EPERM; | |
1075 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1076 | break; | |
1077 | ||
1078 | err = open_related_ns(&net->ns, get_net_ns); | |
1079 | break; | |
1da177e4 | 1080 | default: |
6b96018b | 1081 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1082 | break; |
89bddce5 | 1083 | } |
1da177e4 LT |
1084 | return err; |
1085 | } | |
1086 | ||
1087 | int sock_create_lite(int family, int type, int protocol, struct socket **res) | |
1088 | { | |
1089 | int err; | |
1090 | struct socket *sock = NULL; | |
89bddce5 | 1091 | |
1da177e4 LT |
1092 | err = security_socket_create(family, type, protocol, 1); |
1093 | if (err) | |
1094 | goto out; | |
1095 | ||
1096 | sock = sock_alloc(); | |
1097 | if (!sock) { | |
1098 | err = -ENOMEM; | |
1099 | goto out; | |
1100 | } | |
1101 | ||
1da177e4 | 1102 | sock->type = type; |
7420ed23 VY |
1103 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1104 | if (err) | |
1105 | goto out_release; | |
1106 | ||
1da177e4 LT |
1107 | out: |
1108 | *res = sock; | |
1109 | return err; | |
7420ed23 VY |
1110 | out_release: |
1111 | sock_release(sock); | |
1112 | sock = NULL; | |
1113 | goto out; | |
1da177e4 | 1114 | } |
c6d409cf | 1115 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1116 | |
1117 | /* No kernel lock held - perfect */ | |
ade994f4 | 1118 | static __poll_t sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1119 | { |
e6c8adca | 1120 | __poll_t busy_flag = 0; |
1da177e4 LT |
1121 | struct socket *sock; |
1122 | ||
1123 | /* | |
89bddce5 | 1124 | * We can't return errors to poll, so it's either yes or no. |
1da177e4 | 1125 | */ |
b69aee04 | 1126 | sock = file->private_data; |
2d48d67f | 1127 | |
cbf55001 | 1128 | if (sk_can_busy_loop(sock->sk)) { |
2d48d67f | 1129 | /* this socket can poll_ll so tell the system call */ |
cbf55001 | 1130 | busy_flag = POLL_BUSY_LOOP; |
2d48d67f ET |
1131 | |
1132 | /* once, only if requested by syscall */ | |
cbf55001 ET |
1133 | if (wait && (wait->_key & POLL_BUSY_LOOP)) |
1134 | sk_busy_loop(sock->sk, 1); | |
2d48d67f ET |
1135 | } |
1136 | ||
cbf55001 | 1137 | return busy_flag | sock->ops->poll(file, sock, wait); |
1da177e4 LT |
1138 | } |
1139 | ||
89bddce5 | 1140 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1141 | { |
b69aee04 | 1142 | struct socket *sock = file->private_data; |
1da177e4 LT |
1143 | |
1144 | return sock->ops->mmap(file, sock, vma); | |
1145 | } | |
1146 | ||
20380731 | 1147 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1148 | { |
1da177e4 LT |
1149 | sock_release(SOCKET_I(inode)); |
1150 | return 0; | |
1151 | } | |
1152 | ||
1153 | /* | |
1154 | * Update the socket async list | |
1155 | * | |
1156 | * Fasync_list locking strategy. | |
1157 | * | |
1158 | * 1. fasync_list is modified only under process context socket lock | |
1159 | * i.e. under semaphore. | |
1160 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1161 | * or under socket lock |
1da177e4 LT |
1162 | */ |
1163 | ||
1164 | static int sock_fasync(int fd, struct file *filp, int on) | |
1165 | { | |
989a2979 ED |
1166 | struct socket *sock = filp->private_data; |
1167 | struct sock *sk = sock->sk; | |
eaefd110 | 1168 | struct socket_wq *wq; |
1da177e4 | 1169 | |
989a2979 | 1170 | if (sk == NULL) |
1da177e4 | 1171 | return -EINVAL; |
1da177e4 LT |
1172 | |
1173 | lock_sock(sk); | |
1e1d04e6 | 1174 | wq = rcu_dereference_protected(sock->wq, lockdep_sock_is_held(sk)); |
eaefd110 | 1175 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1176 | |
eaefd110 | 1177 | if (!wq->fasync_list) |
989a2979 ED |
1178 | sock_reset_flag(sk, SOCK_FASYNC); |
1179 | else | |
bcdce719 | 1180 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1181 | |
989a2979 | 1182 | release_sock(sk); |
1da177e4 LT |
1183 | return 0; |
1184 | } | |
1185 | ||
ceb5d58b | 1186 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1187 | |
ceb5d58b | 1188 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1189 | { |
ceb5d58b | 1190 | if (!wq || !wq->fasync_list) |
1da177e4 | 1191 | return -1; |
ceb5d58b | 1192 | |
89bddce5 | 1193 | switch (how) { |
8d8ad9d7 | 1194 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1195 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1196 | break; |
1197 | goto call_kill; | |
8d8ad9d7 | 1198 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1199 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 LT |
1200 | break; |
1201 | /* fall through */ | |
8d8ad9d7 | 1202 | case SOCK_WAKE_IO: |
89bddce5 | 1203 | call_kill: |
43815482 | 1204 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1205 | break; |
8d8ad9d7 | 1206 | case SOCK_WAKE_URG: |
43815482 | 1207 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1208 | } |
ceb5d58b | 1209 | |
1da177e4 LT |
1210 | return 0; |
1211 | } | |
c6d409cf | 1212 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1213 | |
721db93a | 1214 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1215 | struct socket **res, int kern) |
1da177e4 LT |
1216 | { |
1217 | int err; | |
1218 | struct socket *sock; | |
55737fda | 1219 | const struct net_proto_family *pf; |
1da177e4 LT |
1220 | |
1221 | /* | |
89bddce5 | 1222 | * Check protocol is in range |
1da177e4 LT |
1223 | */ |
1224 | if (family < 0 || family >= NPROTO) | |
1225 | return -EAFNOSUPPORT; | |
1226 | if (type < 0 || type >= SOCK_MAX) | |
1227 | return -EINVAL; | |
1228 | ||
1229 | /* Compatibility. | |
1230 | ||
1231 | This uglymoron is moved from INET layer to here to avoid | |
1232 | deadlock in module load. | |
1233 | */ | |
1234 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1235 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1236 | current->comm); | |
1da177e4 LT |
1237 | family = PF_PACKET; |
1238 | } | |
1239 | ||
1240 | err = security_socket_create(family, type, protocol, kern); | |
1241 | if (err) | |
1242 | return err; | |
89bddce5 | 1243 | |
55737fda SH |
1244 | /* |
1245 | * Allocate the socket and allow the family to set things up. if | |
1246 | * the protocol is 0, the family is instructed to select an appropriate | |
1247 | * default. | |
1248 | */ | |
1249 | sock = sock_alloc(); | |
1250 | if (!sock) { | |
e87cc472 | 1251 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1252 | return -ENFILE; /* Not exactly a match, but its the |
1253 | closest posix thing */ | |
1254 | } | |
1255 | ||
1256 | sock->type = type; | |
1257 | ||
95a5afca | 1258 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1259 | /* Attempt to load a protocol module if the find failed. |
1260 | * | |
1261 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1262 | * requested real, full-featured networking support upon configuration. |
1263 | * Otherwise module support will break! | |
1264 | */ | |
190683a9 | 1265 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1266 | request_module("net-pf-%d", family); |
1da177e4 LT |
1267 | #endif |
1268 | ||
55737fda SH |
1269 | rcu_read_lock(); |
1270 | pf = rcu_dereference(net_families[family]); | |
1271 | err = -EAFNOSUPPORT; | |
1272 | if (!pf) | |
1273 | goto out_release; | |
1da177e4 LT |
1274 | |
1275 | /* | |
1276 | * We will call the ->create function, that possibly is in a loadable | |
1277 | * module, so we have to bump that loadable module refcnt first. | |
1278 | */ | |
55737fda | 1279 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1280 | goto out_release; |
1281 | ||
55737fda SH |
1282 | /* Now protected by module ref count */ |
1283 | rcu_read_unlock(); | |
1284 | ||
3f378b68 | 1285 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1286 | if (err < 0) |
1da177e4 | 1287 | goto out_module_put; |
a79af59e | 1288 | |
1da177e4 LT |
1289 | /* |
1290 | * Now to bump the refcnt of the [loadable] module that owns this | |
1291 | * socket at sock_release time we decrement its refcnt. | |
1292 | */ | |
55737fda SH |
1293 | if (!try_module_get(sock->ops->owner)) |
1294 | goto out_module_busy; | |
1295 | ||
1da177e4 LT |
1296 | /* |
1297 | * Now that we're done with the ->create function, the [loadable] | |
1298 | * module can have its refcnt decremented | |
1299 | */ | |
55737fda | 1300 | module_put(pf->owner); |
7420ed23 VY |
1301 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1302 | if (err) | |
3b185525 | 1303 | goto out_sock_release; |
55737fda | 1304 | *res = sock; |
1da177e4 | 1305 | |
55737fda SH |
1306 | return 0; |
1307 | ||
1308 | out_module_busy: | |
1309 | err = -EAFNOSUPPORT; | |
1da177e4 | 1310 | out_module_put: |
55737fda SH |
1311 | sock->ops = NULL; |
1312 | module_put(pf->owner); | |
1313 | out_sock_release: | |
1da177e4 | 1314 | sock_release(sock); |
55737fda SH |
1315 | return err; |
1316 | ||
1317 | out_release: | |
1318 | rcu_read_unlock(); | |
1319 | goto out_sock_release; | |
1da177e4 | 1320 | } |
721db93a | 1321 | EXPORT_SYMBOL(__sock_create); |
1da177e4 LT |
1322 | |
1323 | int sock_create(int family, int type, int protocol, struct socket **res) | |
1324 | { | |
1b8d7ae4 | 1325 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1326 | } |
c6d409cf | 1327 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1328 | |
eeb1bd5c | 1329 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1330 | { |
eeb1bd5c | 1331 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1332 | } |
c6d409cf | 1333 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1334 | |
9d6a15c3 | 1335 | int __sys_socket(int family, int type, int protocol) |
1da177e4 LT |
1336 | { |
1337 | int retval; | |
1338 | struct socket *sock; | |
a677a039 UD |
1339 | int flags; |
1340 | ||
e38b36f3 UD |
1341 | /* Check the SOCK_* constants for consistency. */ |
1342 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1343 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1344 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1345 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1346 | ||
a677a039 | 1347 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1348 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1349 | return -EINVAL; |
1350 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1351 | |
aaca0bdc UD |
1352 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1353 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1354 | ||
1da177e4 LT |
1355 | retval = sock_create(family, type, protocol, &sock); |
1356 | if (retval < 0) | |
8e1611e2 | 1357 | return retval; |
1da177e4 | 1358 | |
8e1611e2 | 1359 | return sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1360 | } |
1361 | ||
9d6a15c3 DB |
1362 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1363 | { | |
1364 | return __sys_socket(family, type, protocol); | |
1365 | } | |
1366 | ||
1da177e4 LT |
1367 | /* |
1368 | * Create a pair of connected sockets. | |
1369 | */ | |
1370 | ||
3e0fa65f HC |
1371 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1372 | int __user *, usockvec) | |
1da177e4 LT |
1373 | { |
1374 | struct socket *sock1, *sock2; | |
1375 | int fd1, fd2, err; | |
db349509 | 1376 | struct file *newfile1, *newfile2; |
a677a039 UD |
1377 | int flags; |
1378 | ||
1379 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1380 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1381 | return -EINVAL; |
1382 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1383 | |
aaca0bdc UD |
1384 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1385 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1386 | ||
016a266b AV |
1387 | /* |
1388 | * reserve descriptors and make sure we won't fail | |
1389 | * to return them to userland. | |
1390 | */ | |
1391 | fd1 = get_unused_fd_flags(flags); | |
1392 | if (unlikely(fd1 < 0)) | |
1393 | return fd1; | |
1394 | ||
1395 | fd2 = get_unused_fd_flags(flags); | |
1396 | if (unlikely(fd2 < 0)) { | |
1397 | put_unused_fd(fd1); | |
1398 | return fd2; | |
1399 | } | |
1400 | ||
1401 | err = put_user(fd1, &usockvec[0]); | |
1402 | if (err) | |
1403 | goto out; | |
1404 | ||
1405 | err = put_user(fd2, &usockvec[1]); | |
1406 | if (err) | |
1407 | goto out; | |
1408 | ||
1da177e4 LT |
1409 | /* |
1410 | * Obtain the first socket and check if the underlying protocol | |
1411 | * supports the socketpair call. | |
1412 | */ | |
1413 | ||
1414 | err = sock_create(family, type, protocol, &sock1); | |
016a266b | 1415 | if (unlikely(err < 0)) |
1da177e4 LT |
1416 | goto out; |
1417 | ||
1418 | err = sock_create(family, type, protocol, &sock2); | |
016a266b AV |
1419 | if (unlikely(err < 0)) { |
1420 | sock_release(sock1); | |
1421 | goto out; | |
bf3c23d1 | 1422 | } |
d73aa286 | 1423 | |
016a266b AV |
1424 | err = sock1->ops->socketpair(sock1, sock2); |
1425 | if (unlikely(err < 0)) { | |
1426 | sock_release(sock2); | |
1427 | sock_release(sock1); | |
1428 | goto out; | |
28407630 AV |
1429 | } |
1430 | ||
aab174f0 | 1431 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1432 | if (IS_ERR(newfile1)) { |
28407630 | 1433 | err = PTR_ERR(newfile1); |
016a266b AV |
1434 | sock_release(sock2); |
1435 | goto out; | |
28407630 AV |
1436 | } |
1437 | ||
aab174f0 | 1438 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1439 | if (IS_ERR(newfile2)) { |
1440 | err = PTR_ERR(newfile2); | |
016a266b AV |
1441 | fput(newfile1); |
1442 | goto out; | |
db349509 AV |
1443 | } |
1444 | ||
157cf649 | 1445 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1446 | |
db349509 AV |
1447 | fd_install(fd1, newfile1); |
1448 | fd_install(fd2, newfile2); | |
d73aa286 | 1449 | return 0; |
1da177e4 | 1450 | |
016a266b | 1451 | out: |
d73aa286 | 1452 | put_unused_fd(fd2); |
d73aa286 | 1453 | put_unused_fd(fd1); |
1da177e4 LT |
1454 | return err; |
1455 | } | |
1456 | ||
1da177e4 LT |
1457 | /* |
1458 | * Bind a name to a socket. Nothing much to do here since it's | |
1459 | * the protocol's responsibility to handle the local address. | |
1460 | * | |
1461 | * We move the socket address to kernel space before we call | |
1462 | * the protocol layer (having also checked the address is ok). | |
1463 | */ | |
1464 | ||
a87d35d8 | 1465 | int __sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
1da177e4 LT |
1466 | { |
1467 | struct socket *sock; | |
230b1839 | 1468 | struct sockaddr_storage address; |
6cb153ca | 1469 | int err, fput_needed; |
1da177e4 | 1470 | |
89bddce5 | 1471 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1472 | if (sock) { |
43db362d | 1473 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
89bddce5 SH |
1474 | if (err >= 0) { |
1475 | err = security_socket_bind(sock, | |
230b1839 | 1476 | (struct sockaddr *)&address, |
89bddce5 | 1477 | addrlen); |
6cb153ca BL |
1478 | if (!err) |
1479 | err = sock->ops->bind(sock, | |
89bddce5 | 1480 | (struct sockaddr *) |
230b1839 | 1481 | &address, addrlen); |
1da177e4 | 1482 | } |
6cb153ca | 1483 | fput_light(sock->file, fput_needed); |
89bddce5 | 1484 | } |
1da177e4 LT |
1485 | return err; |
1486 | } | |
1487 | ||
a87d35d8 DB |
1488 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1489 | { | |
1490 | return __sys_bind(fd, umyaddr, addrlen); | |
1491 | } | |
1492 | ||
1da177e4 LT |
1493 | /* |
1494 | * Perform a listen. Basically, we allow the protocol to do anything | |
1495 | * necessary for a listen, and if that works, we mark the socket as | |
1496 | * ready for listening. | |
1497 | */ | |
1498 | ||
3e0fa65f | 1499 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1da177e4 LT |
1500 | { |
1501 | struct socket *sock; | |
6cb153ca | 1502 | int err, fput_needed; |
b8e1f9b5 | 1503 | int somaxconn; |
89bddce5 SH |
1504 | |
1505 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1506 | if (sock) { | |
8efa6e93 | 1507 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1508 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1509 | backlog = somaxconn; |
1da177e4 LT |
1510 | |
1511 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1512 | if (!err) |
1513 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1514 | |
6cb153ca | 1515 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1516 | } |
1517 | return err; | |
1518 | } | |
1519 | ||
1da177e4 LT |
1520 | /* |
1521 | * For accept, we attempt to create a new socket, set up the link | |
1522 | * with the client, wake up the client, then return the new | |
1523 | * connected fd. We collect the address of the connector in kernel | |
1524 | * space and move it to user at the very end. This is unclean because | |
1525 | * we open the socket then return an error. | |
1526 | * | |
1527 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1528 | * status to recvmsg. We need to add that support in a way thats | |
1529 | * clean when we restucture accept also. | |
1530 | */ | |
1531 | ||
4541e805 DB |
1532 | int __sys_accept4(int fd, struct sockaddr __user *upeer_sockaddr, |
1533 | int __user *upeer_addrlen, int flags) | |
1da177e4 LT |
1534 | { |
1535 | struct socket *sock, *newsock; | |
39d8c1b6 | 1536 | struct file *newfile; |
6cb153ca | 1537 | int err, len, newfd, fput_needed; |
230b1839 | 1538 | struct sockaddr_storage address; |
1da177e4 | 1539 | |
77d27200 | 1540 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1541 | return -EINVAL; |
1542 | ||
1543 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1544 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1545 | ||
6cb153ca | 1546 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1547 | if (!sock) |
1548 | goto out; | |
1549 | ||
1550 | err = -ENFILE; | |
c6d409cf ED |
1551 | newsock = sock_alloc(); |
1552 | if (!newsock) | |
1da177e4 LT |
1553 | goto out_put; |
1554 | ||
1555 | newsock->type = sock->type; | |
1556 | newsock->ops = sock->ops; | |
1557 | ||
1da177e4 LT |
1558 | /* |
1559 | * We don't need try_module_get here, as the listening socket (sock) | |
1560 | * has the protocol module (sock->ops->owner) held. | |
1561 | */ | |
1562 | __module_get(newsock->ops->owner); | |
1563 | ||
28407630 | 1564 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1565 | if (unlikely(newfd < 0)) { |
1566 | err = newfd; | |
9a1875e6 DM |
1567 | sock_release(newsock); |
1568 | goto out_put; | |
39d8c1b6 | 1569 | } |
aab174f0 | 1570 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1571 | if (IS_ERR(newfile)) { |
28407630 AV |
1572 | err = PTR_ERR(newfile); |
1573 | put_unused_fd(newfd); | |
28407630 AV |
1574 | goto out_put; |
1575 | } | |
39d8c1b6 | 1576 | |
a79af59e FF |
1577 | err = security_socket_accept(sock, newsock); |
1578 | if (err) | |
39d8c1b6 | 1579 | goto out_fd; |
a79af59e | 1580 | |
cdfbabfb | 1581 | err = sock->ops->accept(sock, newsock, sock->file->f_flags, false); |
1da177e4 | 1582 | if (err < 0) |
39d8c1b6 | 1583 | goto out_fd; |
1da177e4 LT |
1584 | |
1585 | if (upeer_sockaddr) { | |
230b1839 | 1586 | if (newsock->ops->getname(newsock, (struct sockaddr *)&address, |
89bddce5 | 1587 | &len, 2) < 0) { |
1da177e4 | 1588 | err = -ECONNABORTED; |
39d8c1b6 | 1589 | goto out_fd; |
1da177e4 | 1590 | } |
43db362d | 1591 | err = move_addr_to_user(&address, |
230b1839 | 1592 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1593 | if (err < 0) |
39d8c1b6 | 1594 | goto out_fd; |
1da177e4 LT |
1595 | } |
1596 | ||
1597 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1598 | ||
39d8c1b6 DM |
1599 | fd_install(newfd, newfile); |
1600 | err = newfd; | |
1da177e4 | 1601 | |
1da177e4 | 1602 | out_put: |
6cb153ca | 1603 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1604 | out: |
1605 | return err; | |
39d8c1b6 | 1606 | out_fd: |
9606a216 | 1607 | fput(newfile); |
39d8c1b6 | 1608 | put_unused_fd(newfd); |
1da177e4 LT |
1609 | goto out_put; |
1610 | } | |
1611 | ||
4541e805 DB |
1612 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1613 | int __user *, upeer_addrlen, int, flags) | |
1614 | { | |
1615 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, flags); | |
1616 | } | |
1617 | ||
20f37034 HC |
1618 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1619 | int __user *, upeer_addrlen) | |
aaca0bdc | 1620 | { |
4541e805 | 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 | ||
1387c2c2 | 1636 | int __sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
1da177e4 LT |
1637 | { |
1638 | struct socket *sock; | |
230b1839 | 1639 | struct sockaddr_storage address; |
6cb153ca | 1640 | int err, fput_needed; |
1da177e4 | 1641 | |
6cb153ca | 1642 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1643 | if (!sock) |
1644 | goto out; | |
43db362d | 1645 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1646 | if (err < 0) |
1647 | goto out_put; | |
1648 | ||
89bddce5 | 1649 | err = |
230b1839 | 1650 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1651 | if (err) |
1652 | goto out_put; | |
1653 | ||
230b1839 | 1654 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1655 | sock->file->f_flags); |
1656 | out_put: | |
6cb153ca | 1657 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1658 | out: |
1659 | return err; | |
1660 | } | |
1661 | ||
1387c2c2 DB |
1662 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1663 | int, addrlen) | |
1664 | { | |
1665 | return __sys_connect(fd, uservaddr, addrlen); | |
1666 | } | |
1667 | ||
1da177e4 LT |
1668 | /* |
1669 | * Get the local address ('name') of a socket object. Move the obtained | |
1670 | * name to user space. | |
1671 | */ | |
1672 | ||
20f37034 HC |
1673 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1674 | int __user *, usockaddr_len) | |
1da177e4 LT |
1675 | { |
1676 | struct socket *sock; | |
230b1839 | 1677 | struct sockaddr_storage address; |
6cb153ca | 1678 | int len, err, fput_needed; |
89bddce5 | 1679 | |
6cb153ca | 1680 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1681 | if (!sock) |
1682 | goto out; | |
1683 | ||
1684 | err = security_socket_getsockname(sock); | |
1685 | if (err) | |
1686 | goto out_put; | |
1687 | ||
230b1839 | 1688 | err = sock->ops->getname(sock, (struct sockaddr *)&address, &len, 0); |
1da177e4 LT |
1689 | if (err) |
1690 | goto out_put; | |
43db362d | 1691 | err = move_addr_to_user(&address, len, usockaddr, usockaddr_len); |
1da177e4 LT |
1692 | |
1693 | out_put: | |
6cb153ca | 1694 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1695 | out: |
1696 | return err; | |
1697 | } | |
1698 | ||
1699 | /* | |
1700 | * Get the remote address ('name') of a socket object. Move the obtained | |
1701 | * name to user space. | |
1702 | */ | |
1703 | ||
20f37034 HC |
1704 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1705 | int __user *, usockaddr_len) | |
1da177e4 LT |
1706 | { |
1707 | struct socket *sock; | |
230b1839 | 1708 | struct sockaddr_storage address; |
6cb153ca | 1709 | int len, err, fput_needed; |
1da177e4 | 1710 | |
89bddce5 SH |
1711 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1712 | if (sock != NULL) { | |
1da177e4 LT |
1713 | err = security_socket_getpeername(sock); |
1714 | if (err) { | |
6cb153ca | 1715 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1716 | return err; |
1717 | } | |
1718 | ||
89bddce5 | 1719 | err = |
230b1839 | 1720 | sock->ops->getname(sock, (struct sockaddr *)&address, &len, |
89bddce5 | 1721 | 1); |
1da177e4 | 1722 | if (!err) |
43db362d | 1723 | err = move_addr_to_user(&address, len, usockaddr, |
89bddce5 | 1724 | usockaddr_len); |
6cb153ca | 1725 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1726 | } |
1727 | return err; | |
1728 | } | |
1729 | ||
1730 | /* | |
1731 | * Send a datagram to a given address. We move the address into kernel | |
1732 | * space and check the user space data area is readable before invoking | |
1733 | * the protocol. | |
1734 | */ | |
211b634b DB |
1735 | int __sys_sendto(int fd, void __user *buff, size_t len, unsigned int flags, |
1736 | struct sockaddr __user *addr, int addr_len) | |
1da177e4 LT |
1737 | { |
1738 | struct socket *sock; | |
230b1839 | 1739 | struct sockaddr_storage address; |
1da177e4 LT |
1740 | int err; |
1741 | struct msghdr msg; | |
1742 | struct iovec iov; | |
6cb153ca | 1743 | int fput_needed; |
6cb153ca | 1744 | |
602bd0e9 AV |
1745 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1746 | if (unlikely(err)) | |
1747 | return err; | |
de0fa95c PE |
1748 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1749 | if (!sock) | |
4387ff75 | 1750 | goto out; |
6cb153ca | 1751 | |
89bddce5 | 1752 | msg.msg_name = NULL; |
89bddce5 SH |
1753 | msg.msg_control = NULL; |
1754 | msg.msg_controllen = 0; | |
1755 | msg.msg_namelen = 0; | |
6cb153ca | 1756 | if (addr) { |
43db362d | 1757 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1758 | if (err < 0) |
1759 | goto out_put; | |
230b1839 | 1760 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1761 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1762 | } |
1763 | if (sock->file->f_flags & O_NONBLOCK) | |
1764 | flags |= MSG_DONTWAIT; | |
1765 | msg.msg_flags = flags; | |
d8725c86 | 1766 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 1767 | |
89bddce5 | 1768 | out_put: |
de0fa95c | 1769 | fput_light(sock->file, fput_needed); |
4387ff75 | 1770 | out: |
1da177e4 LT |
1771 | return err; |
1772 | } | |
1773 | ||
211b634b DB |
1774 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
1775 | unsigned int, flags, struct sockaddr __user *, addr, | |
1776 | int, addr_len) | |
1777 | { | |
1778 | return __sys_sendto(fd, buff, len, flags, addr, addr_len); | |
1779 | } | |
1780 | ||
1da177e4 | 1781 | /* |
89bddce5 | 1782 | * Send a datagram down a socket. |
1da177e4 LT |
1783 | */ |
1784 | ||
3e0fa65f | 1785 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1786 | unsigned int, flags) |
1da177e4 | 1787 | { |
211b634b | 1788 | return __sys_sendto(fd, buff, len, flags, NULL, 0); |
1da177e4 LT |
1789 | } |
1790 | ||
1791 | /* | |
89bddce5 | 1792 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1793 | * sender. We verify the buffers are writable and if needed move the |
1794 | * sender address from kernel to user space. | |
1795 | */ | |
7a09e1eb DB |
1796 | int __sys_recvfrom(int fd, void __user *ubuf, size_t size, unsigned int flags, |
1797 | struct sockaddr __user *addr, int __user *addr_len) | |
1da177e4 LT |
1798 | { |
1799 | struct socket *sock; | |
1800 | struct iovec iov; | |
1801 | struct msghdr msg; | |
230b1839 | 1802 | struct sockaddr_storage address; |
89bddce5 | 1803 | int err, err2; |
6cb153ca BL |
1804 | int fput_needed; |
1805 | ||
602bd0e9 AV |
1806 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
1807 | if (unlikely(err)) | |
1808 | return err; | |
de0fa95c | 1809 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 1810 | if (!sock) |
de0fa95c | 1811 | goto out; |
1da177e4 | 1812 | |
89bddce5 SH |
1813 | msg.msg_control = NULL; |
1814 | msg.msg_controllen = 0; | |
f3d33426 HFS |
1815 | /* Save some cycles and don't copy the address if not needed */ |
1816 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
1817 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
1818 | msg.msg_namelen = 0; | |
130ed5d1 | 1819 | msg.msg_iocb = NULL; |
9f138fa6 | 1820 | msg.msg_flags = 0; |
1da177e4 LT |
1821 | if (sock->file->f_flags & O_NONBLOCK) |
1822 | flags |= MSG_DONTWAIT; | |
2da62906 | 1823 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 1824 | |
89bddce5 | 1825 | if (err >= 0 && addr != NULL) { |
43db362d | 1826 | err2 = move_addr_to_user(&address, |
230b1839 | 1827 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
1828 | if (err2 < 0) |
1829 | err = err2; | |
1da177e4 | 1830 | } |
de0fa95c PE |
1831 | |
1832 | fput_light(sock->file, fput_needed); | |
4387ff75 | 1833 | out: |
1da177e4 LT |
1834 | return err; |
1835 | } | |
1836 | ||
7a09e1eb DB |
1837 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
1838 | unsigned int, flags, struct sockaddr __user *, addr, | |
1839 | int __user *, addr_len) | |
1840 | { | |
1841 | return __sys_recvfrom(fd, ubuf, size, flags, addr, addr_len); | |
1842 | } | |
1843 | ||
1da177e4 | 1844 | /* |
89bddce5 | 1845 | * Receive a datagram from a socket. |
1da177e4 LT |
1846 | */ |
1847 | ||
b7c0ddf5 JG |
1848 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
1849 | unsigned int, flags) | |
1da177e4 | 1850 | { |
7a09e1eb | 1851 | return __sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
1da177e4 LT |
1852 | } |
1853 | ||
1854 | /* | |
1855 | * Set a socket option. Because we don't know the option lengths we have | |
1856 | * to pass the user mode parameter for the protocols to sort out. | |
1857 | */ | |
1858 | ||
20f37034 HC |
1859 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
1860 | char __user *, optval, int, optlen) | |
1da177e4 | 1861 | { |
6cb153ca | 1862 | int err, fput_needed; |
1da177e4 LT |
1863 | struct socket *sock; |
1864 | ||
1865 | if (optlen < 0) | |
1866 | return -EINVAL; | |
89bddce5 SH |
1867 | |
1868 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1869 | if (sock != NULL) { | |
1870 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
1871 | if (err) |
1872 | goto out_put; | |
1da177e4 LT |
1873 | |
1874 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1875 | err = |
1876 | sock_setsockopt(sock, level, optname, optval, | |
1877 | optlen); | |
1da177e4 | 1878 | else |
89bddce5 SH |
1879 | err = |
1880 | sock->ops->setsockopt(sock, level, optname, optval, | |
1881 | optlen); | |
6cb153ca BL |
1882 | out_put: |
1883 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1884 | } |
1885 | return err; | |
1886 | } | |
1887 | ||
1888 | /* | |
1889 | * Get a socket option. Because we don't know the option lengths we have | |
1890 | * to pass a user mode parameter for the protocols to sort out. | |
1891 | */ | |
1892 | ||
20f37034 HC |
1893 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
1894 | char __user *, optval, int __user *, optlen) | |
1da177e4 | 1895 | { |
6cb153ca | 1896 | int err, fput_needed; |
1da177e4 LT |
1897 | struct socket *sock; |
1898 | ||
89bddce5 SH |
1899 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1900 | if (sock != NULL) { | |
6cb153ca BL |
1901 | err = security_socket_getsockopt(sock, level, optname); |
1902 | if (err) | |
1903 | goto out_put; | |
1da177e4 LT |
1904 | |
1905 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1906 | err = |
1907 | sock_getsockopt(sock, level, optname, optval, | |
1908 | optlen); | |
1da177e4 | 1909 | else |
89bddce5 SH |
1910 | err = |
1911 | sock->ops->getsockopt(sock, level, optname, optval, | |
1912 | optlen); | |
6cb153ca BL |
1913 | out_put: |
1914 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1915 | } |
1916 | return err; | |
1917 | } | |
1918 | ||
1da177e4 LT |
1919 | /* |
1920 | * Shutdown a socket. | |
1921 | */ | |
1922 | ||
754fe8d2 | 1923 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
1da177e4 | 1924 | { |
6cb153ca | 1925 | int err, fput_needed; |
1da177e4 LT |
1926 | struct socket *sock; |
1927 | ||
89bddce5 SH |
1928 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1929 | if (sock != NULL) { | |
1da177e4 | 1930 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
1931 | if (!err) |
1932 | err = sock->ops->shutdown(sock, how); | |
1933 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1934 | } |
1935 | return err; | |
1936 | } | |
1937 | ||
89bddce5 | 1938 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
1939 | * fields which are the same type (int / unsigned) on our platforms. |
1940 | */ | |
1941 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
1942 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
1943 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
1944 | ||
c71d8ebe TH |
1945 | struct used_address { |
1946 | struct sockaddr_storage name; | |
1947 | unsigned int name_len; | |
1948 | }; | |
1949 | ||
da184284 AV |
1950 | static int copy_msghdr_from_user(struct msghdr *kmsg, |
1951 | struct user_msghdr __user *umsg, | |
1952 | struct sockaddr __user **save_addr, | |
1953 | struct iovec **iov) | |
1661bf36 | 1954 | { |
ffb07550 | 1955 | struct user_msghdr msg; |
08adb7da AV |
1956 | ssize_t err; |
1957 | ||
ffb07550 | 1958 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 1959 | return -EFAULT; |
dbb490b9 | 1960 | |
864d9664 | 1961 | kmsg->msg_control = (void __force *)msg.msg_control; |
ffb07550 AV |
1962 | kmsg->msg_controllen = msg.msg_controllen; |
1963 | kmsg->msg_flags = msg.msg_flags; | |
1964 | ||
1965 | kmsg->msg_namelen = msg.msg_namelen; | |
1966 | if (!msg.msg_name) | |
6a2a2b3a AS |
1967 | kmsg->msg_namelen = 0; |
1968 | ||
dbb490b9 ML |
1969 | if (kmsg->msg_namelen < 0) |
1970 | return -EINVAL; | |
1971 | ||
1661bf36 | 1972 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 1973 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
1974 | |
1975 | if (save_addr) | |
ffb07550 | 1976 | *save_addr = msg.msg_name; |
08adb7da | 1977 | |
ffb07550 | 1978 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 1979 | if (!save_addr) { |
864d9664 PA |
1980 | err = move_addr_to_kernel(msg.msg_name, |
1981 | kmsg->msg_namelen, | |
08adb7da AV |
1982 | kmsg->msg_name); |
1983 | if (err < 0) | |
1984 | return err; | |
1985 | } | |
1986 | } else { | |
1987 | kmsg->msg_name = NULL; | |
1988 | kmsg->msg_namelen = 0; | |
1989 | } | |
1990 | ||
ffb07550 | 1991 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
1992 | return -EMSGSIZE; |
1993 | ||
0345f931 | 1994 | kmsg->msg_iocb = NULL; |
1995 | ||
ffb07550 AV |
1996 | return import_iovec(save_addr ? READ : WRITE, |
1997 | msg.msg_iov, msg.msg_iovlen, | |
da184284 | 1998 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
1661bf36 DC |
1999 | } |
2000 | ||
666547ff | 2001 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2002 | struct msghdr *msg_sys, unsigned int flags, |
28a94d8f TH |
2003 | struct used_address *used_address, |
2004 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 2005 | { |
89bddce5 SH |
2006 | struct compat_msghdr __user *msg_compat = |
2007 | (struct compat_msghdr __user *)msg; | |
230b1839 | 2008 | struct sockaddr_storage address; |
1da177e4 | 2009 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 2010 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 2011 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 2012 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 2013 | unsigned char *ctl_buf = ctl; |
d8725c86 | 2014 | int ctl_len; |
08adb7da | 2015 | ssize_t err; |
89bddce5 | 2016 | |
08adb7da | 2017 | msg_sys->msg_name = &address; |
1da177e4 | 2018 | |
08449320 | 2019 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2020 | err = get_compat_msghdr(msg_sys, msg_compat, NULL, &iov); |
08449320 | 2021 | else |
08adb7da | 2022 | err = copy_msghdr_from_user(msg_sys, msg, NULL, &iov); |
89bddce5 | 2023 | if (err < 0) |
da184284 | 2024 | return err; |
1da177e4 LT |
2025 | |
2026 | err = -ENOBUFS; | |
2027 | ||
228e548e | 2028 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 2029 | goto out_freeiov; |
28a94d8f | 2030 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2031 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2032 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2033 | err = |
228e548e | 2034 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2035 | sizeof(ctl)); |
1da177e4 LT |
2036 | if (err) |
2037 | goto out_freeiov; | |
228e548e AB |
2038 | ctl_buf = msg_sys->msg_control; |
2039 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2040 | } else if (ctl_len) { |
ac4340fc DM |
2041 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2042 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2043 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2044 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2045 | if (ctl_buf == NULL) |
1da177e4 LT |
2046 | goto out_freeiov; |
2047 | } | |
2048 | err = -EFAULT; | |
2049 | /* | |
228e548e | 2050 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2051 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2052 | * checking falls down on this. | |
2053 | */ | |
fb8621bb | 2054 | if (copy_from_user(ctl_buf, |
228e548e | 2055 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2056 | ctl_len)) |
1da177e4 | 2057 | goto out_freectl; |
228e548e | 2058 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2059 | } |
228e548e | 2060 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2061 | |
2062 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2063 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2064 | /* |
2065 | * If this is sendmmsg() and current destination address is same as | |
2066 | * previously succeeded address, omit asking LSM's decision. | |
2067 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2068 | * destination address never matches. | |
2069 | */ | |
bc909d9d MD |
2070 | if (used_address && msg_sys->msg_name && |
2071 | used_address->name_len == msg_sys->msg_namelen && | |
2072 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2073 | used_address->name_len)) { |
d8725c86 | 2074 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2075 | goto out_freectl; |
2076 | } | |
d8725c86 | 2077 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2078 | /* |
2079 | * If this is sendmmsg() and sending to current destination address was | |
2080 | * successful, remember it. | |
2081 | */ | |
2082 | if (used_address && err >= 0) { | |
2083 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2084 | if (msg_sys->msg_name) |
2085 | memcpy(&used_address->name, msg_sys->msg_name, | |
2086 | used_address->name_len); | |
c71d8ebe | 2087 | } |
1da177e4 LT |
2088 | |
2089 | out_freectl: | |
89bddce5 | 2090 | if (ctl_buf != ctl) |
1da177e4 LT |
2091 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
2092 | out_freeiov: | |
da184284 | 2093 | kfree(iov); |
228e548e AB |
2094 | return err; |
2095 | } | |
2096 | ||
2097 | /* | |
2098 | * BSD sendmsg interface | |
2099 | */ | |
2100 | ||
666547ff | 2101 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned flags) |
228e548e AB |
2102 | { |
2103 | int fput_needed, err; | |
2104 | struct msghdr msg_sys; | |
1be374a0 AL |
2105 | struct socket *sock; |
2106 | ||
1be374a0 | 2107 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2108 | if (!sock) |
2109 | goto out; | |
2110 | ||
28a94d8f | 2111 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2112 | |
6cb153ca | 2113 | fput_light(sock->file, fput_needed); |
89bddce5 | 2114 | out: |
1da177e4 LT |
2115 | return err; |
2116 | } | |
2117 | ||
666547ff | 2118 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 AL |
2119 | { |
2120 | if (flags & MSG_CMSG_COMPAT) | |
2121 | return -EINVAL; | |
2122 | return __sys_sendmsg(fd, msg, flags); | |
2123 | } | |
2124 | ||
228e548e AB |
2125 | /* |
2126 | * Linux sendmmsg interface | |
2127 | */ | |
2128 | ||
2129 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2130 | unsigned int flags) | |
2131 | { | |
2132 | int fput_needed, err, datagrams; | |
2133 | struct socket *sock; | |
2134 | struct mmsghdr __user *entry; | |
2135 | struct compat_mmsghdr __user *compat_entry; | |
2136 | struct msghdr msg_sys; | |
c71d8ebe | 2137 | struct used_address used_address; |
f092276d | 2138 | unsigned int oflags = flags; |
228e548e | 2139 | |
98382f41 AB |
2140 | if (vlen > UIO_MAXIOV) |
2141 | vlen = UIO_MAXIOV; | |
228e548e AB |
2142 | |
2143 | datagrams = 0; | |
2144 | ||
2145 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2146 | if (!sock) | |
2147 | return err; | |
2148 | ||
c71d8ebe | 2149 | used_address.name_len = UINT_MAX; |
228e548e AB |
2150 | entry = mmsg; |
2151 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2152 | err = 0; |
f092276d | 2153 | flags |= MSG_BATCH; |
228e548e AB |
2154 | |
2155 | while (datagrams < vlen) { | |
f092276d TH |
2156 | if (datagrams == vlen - 1) |
2157 | flags = oflags; | |
2158 | ||
228e548e | 2159 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2160 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2161 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2162 | if (err < 0) |
2163 | break; | |
2164 | err = __put_user(err, &compat_entry->msg_len); | |
2165 | ++compat_entry; | |
2166 | } else { | |
a7526eb5 | 2167 | err = ___sys_sendmsg(sock, |
666547ff | 2168 | (struct user_msghdr __user *)entry, |
28a94d8f | 2169 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2170 | if (err < 0) |
2171 | break; | |
2172 | err = put_user(err, &entry->msg_len); | |
2173 | ++entry; | |
2174 | } | |
2175 | ||
2176 | if (err) | |
2177 | break; | |
2178 | ++datagrams; | |
3023898b SHY |
2179 | if (msg_data_left(&msg_sys)) |
2180 | break; | |
a78cb84c | 2181 | cond_resched(); |
228e548e AB |
2182 | } |
2183 | ||
228e548e AB |
2184 | fput_light(sock->file, fput_needed); |
2185 | ||
728ffb86 AB |
2186 | /* We only return an error if no datagrams were able to be sent */ |
2187 | if (datagrams != 0) | |
228e548e AB |
2188 | return datagrams; |
2189 | ||
228e548e AB |
2190 | return err; |
2191 | } | |
2192 | ||
2193 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2194 | unsigned int, vlen, unsigned int, flags) | |
2195 | { | |
1be374a0 AL |
2196 | if (flags & MSG_CMSG_COMPAT) |
2197 | return -EINVAL; | |
228e548e AB |
2198 | return __sys_sendmmsg(fd, mmsg, vlen, flags); |
2199 | } | |
2200 | ||
666547ff | 2201 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2202 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2203 | { |
89bddce5 SH |
2204 | struct compat_msghdr __user *msg_compat = |
2205 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2206 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2207 | struct iovec *iov = iovstack; |
1da177e4 | 2208 | unsigned long cmsg_ptr; |
2da62906 | 2209 | int len; |
08adb7da | 2210 | ssize_t err; |
1da177e4 LT |
2211 | |
2212 | /* kernel mode address */ | |
230b1839 | 2213 | struct sockaddr_storage addr; |
1da177e4 LT |
2214 | |
2215 | /* user mode address pointers */ | |
2216 | struct sockaddr __user *uaddr; | |
08adb7da | 2217 | int __user *uaddr_len = COMPAT_NAMELEN(msg); |
89bddce5 | 2218 | |
08adb7da | 2219 | msg_sys->msg_name = &addr; |
1da177e4 | 2220 | |
f3d33426 | 2221 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2222 | err = get_compat_msghdr(msg_sys, msg_compat, &uaddr, &iov); |
f3d33426 | 2223 | else |
08adb7da | 2224 | err = copy_msghdr_from_user(msg_sys, msg, &uaddr, &iov); |
1da177e4 | 2225 | if (err < 0) |
da184284 | 2226 | return err; |
1da177e4 | 2227 | |
a2e27255 ACM |
2228 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2229 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2230 | |
f3d33426 HFS |
2231 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2232 | msg_sys->msg_namelen = 0; | |
2233 | ||
1da177e4 LT |
2234 | if (sock->file->f_flags & O_NONBLOCK) |
2235 | flags |= MSG_DONTWAIT; | |
2da62906 | 2236 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, flags); |
1da177e4 LT |
2237 | if (err < 0) |
2238 | goto out_freeiov; | |
2239 | len = err; | |
2240 | ||
2241 | if (uaddr != NULL) { | |
43db362d | 2242 | err = move_addr_to_user(&addr, |
a2e27255 | 2243 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2244 | uaddr_len); |
1da177e4 LT |
2245 | if (err < 0) |
2246 | goto out_freeiov; | |
2247 | } | |
a2e27255 | 2248 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2249 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2250 | if (err) |
2251 | goto out_freeiov; | |
2252 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2253 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2254 | &msg_compat->msg_controllen); |
2255 | else | |
a2e27255 | 2256 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2257 | &msg->msg_controllen); |
2258 | if (err) | |
2259 | goto out_freeiov; | |
2260 | err = len; | |
2261 | ||
2262 | out_freeiov: | |
da184284 | 2263 | kfree(iov); |
a2e27255 ACM |
2264 | return err; |
2265 | } | |
2266 | ||
2267 | /* | |
2268 | * BSD recvmsg interface | |
2269 | */ | |
2270 | ||
666547ff | 2271 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned flags) |
a2e27255 ACM |
2272 | { |
2273 | int fput_needed, err; | |
2274 | struct msghdr msg_sys; | |
1be374a0 AL |
2275 | struct socket *sock; |
2276 | ||
1be374a0 | 2277 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2278 | if (!sock) |
2279 | goto out; | |
2280 | ||
a7526eb5 | 2281 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2282 | |
6cb153ca | 2283 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2284 | out: |
2285 | return err; | |
2286 | } | |
2287 | ||
666547ff | 2288 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2289 | unsigned int, flags) |
2290 | { | |
2291 | if (flags & MSG_CMSG_COMPAT) | |
2292 | return -EINVAL; | |
2293 | return __sys_recvmsg(fd, msg, flags); | |
2294 | } | |
2295 | ||
a2e27255 ACM |
2296 | /* |
2297 | * Linux recvmmsg interface | |
2298 | */ | |
2299 | ||
2300 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2301 | unsigned int flags, struct timespec *timeout) | |
2302 | { | |
2303 | int fput_needed, err, datagrams; | |
2304 | struct socket *sock; | |
2305 | struct mmsghdr __user *entry; | |
d7256d0e | 2306 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2307 | struct msghdr msg_sys; |
766b9f92 DD |
2308 | struct timespec64 end_time; |
2309 | struct timespec64 timeout64; | |
a2e27255 ACM |
2310 | |
2311 | if (timeout && | |
2312 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2313 | timeout->tv_nsec)) | |
2314 | return -EINVAL; | |
2315 | ||
2316 | datagrams = 0; | |
2317 | ||
2318 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2319 | if (!sock) | |
2320 | return err; | |
2321 | ||
2322 | err = sock_error(sock->sk); | |
e623a9e9 MJ |
2323 | if (err) { |
2324 | datagrams = err; | |
a2e27255 | 2325 | goto out_put; |
e623a9e9 | 2326 | } |
a2e27255 ACM |
2327 | |
2328 | entry = mmsg; | |
d7256d0e | 2329 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2330 | |
2331 | while (datagrams < vlen) { | |
2332 | /* | |
2333 | * No need to ask LSM for more than the first datagram. | |
2334 | */ | |
d7256d0e | 2335 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2336 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2337 | &msg_sys, flags & ~MSG_WAITFORONE, |
2338 | datagrams); | |
d7256d0e JMG |
2339 | if (err < 0) |
2340 | break; | |
2341 | err = __put_user(err, &compat_entry->msg_len); | |
2342 | ++compat_entry; | |
2343 | } else { | |
a7526eb5 | 2344 | err = ___sys_recvmsg(sock, |
666547ff | 2345 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2346 | &msg_sys, flags & ~MSG_WAITFORONE, |
2347 | datagrams); | |
d7256d0e JMG |
2348 | if (err < 0) |
2349 | break; | |
2350 | err = put_user(err, &entry->msg_len); | |
2351 | ++entry; | |
2352 | } | |
2353 | ||
a2e27255 ACM |
2354 | if (err) |
2355 | break; | |
a2e27255 ACM |
2356 | ++datagrams; |
2357 | ||
71c5c159 BB |
2358 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2359 | if (flags & MSG_WAITFORONE) | |
2360 | flags |= MSG_DONTWAIT; | |
2361 | ||
a2e27255 | 2362 | if (timeout) { |
766b9f92 DD |
2363 | ktime_get_ts64(&timeout64); |
2364 | *timeout = timespec64_to_timespec( | |
2365 | timespec64_sub(end_time, timeout64)); | |
a2e27255 ACM |
2366 | if (timeout->tv_sec < 0) { |
2367 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2368 | break; | |
2369 | } | |
2370 | ||
2371 | /* Timeout, return less than vlen datagrams */ | |
2372 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2373 | break; | |
2374 | } | |
2375 | ||
2376 | /* Out of band data, return right away */ | |
2377 | if (msg_sys.msg_flags & MSG_OOB) | |
2378 | break; | |
a78cb84c | 2379 | cond_resched(); |
a2e27255 ACM |
2380 | } |
2381 | ||
a2e27255 | 2382 | if (err == 0) |
34b88a68 ACM |
2383 | goto out_put; |
2384 | ||
2385 | if (datagrams == 0) { | |
2386 | datagrams = err; | |
2387 | goto out_put; | |
2388 | } | |
a2e27255 | 2389 | |
34b88a68 ACM |
2390 | /* |
2391 | * We may return less entries than requested (vlen) if the | |
2392 | * sock is non block and there aren't enough datagrams... | |
2393 | */ | |
2394 | if (err != -EAGAIN) { | |
a2e27255 | 2395 | /* |
34b88a68 ACM |
2396 | * ... or if recvmsg returns an error after we |
2397 | * received some datagrams, where we record the | |
2398 | * error to return on the next call or if the | |
2399 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2400 | */ |
34b88a68 | 2401 | sock->sk->sk_err = -err; |
a2e27255 | 2402 | } |
34b88a68 ACM |
2403 | out_put: |
2404 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2405 | |
34b88a68 | 2406 | return datagrams; |
a2e27255 ACM |
2407 | } |
2408 | ||
2409 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2410 | unsigned int, vlen, unsigned int, flags, | |
2411 | struct timespec __user *, timeout) | |
2412 | { | |
2413 | int datagrams; | |
2414 | struct timespec timeout_sys; | |
2415 | ||
1be374a0 AL |
2416 | if (flags & MSG_CMSG_COMPAT) |
2417 | return -EINVAL; | |
2418 | ||
a2e27255 ACM |
2419 | if (!timeout) |
2420 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2421 | ||
2422 | if (copy_from_user(&timeout_sys, timeout, sizeof(timeout_sys))) | |
2423 | return -EFAULT; | |
2424 | ||
2425 | datagrams = __sys_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
2426 | ||
2427 | if (datagrams > 0 && | |
2428 | copy_to_user(timeout, &timeout_sys, sizeof(timeout_sys))) | |
2429 | datagrams = -EFAULT; | |
2430 | ||
2431 | return datagrams; | |
2432 | } | |
2433 | ||
2434 | #ifdef __ARCH_WANT_SYS_SOCKETCALL | |
1da177e4 LT |
2435 | /* Argument list sizes for sys_socketcall */ |
2436 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2437 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2438 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2439 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2440 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2441 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2442 | }; |
2443 | ||
1da177e4 LT |
2444 | #undef AL |
2445 | ||
2446 | /* | |
89bddce5 | 2447 | * System call vectors. |
1da177e4 LT |
2448 | * |
2449 | * Argument checking cleaned up. Saved 20% in size. | |
2450 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2451 | * it is set by the callees. |
1da177e4 LT |
2452 | */ |
2453 | ||
3e0fa65f | 2454 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2455 | { |
2950fa9d | 2456 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2457 | unsigned long a0, a1; |
1da177e4 | 2458 | int err; |
47379052 | 2459 | unsigned int len; |
1da177e4 | 2460 | |
228e548e | 2461 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 LT |
2462 | return -EINVAL; |
2463 | ||
47379052 AV |
2464 | len = nargs[call]; |
2465 | if (len > sizeof(a)) | |
2466 | return -EINVAL; | |
2467 | ||
1da177e4 | 2468 | /* copy_from_user should be SMP safe. */ |
47379052 | 2469 | if (copy_from_user(a, args, len)) |
1da177e4 | 2470 | return -EFAULT; |
3ec3b2fb | 2471 | |
2950fa9d CG |
2472 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2473 | if (err) | |
2474 | return err; | |
3ec3b2fb | 2475 | |
89bddce5 SH |
2476 | a0 = a[0]; |
2477 | a1 = a[1]; | |
2478 | ||
2479 | switch (call) { | |
2480 | case SYS_SOCKET: | |
9d6a15c3 | 2481 | err = __sys_socket(a0, a1, a[2]); |
89bddce5 SH |
2482 | break; |
2483 | case SYS_BIND: | |
a87d35d8 | 2484 | err = __sys_bind(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2485 | break; |
2486 | case SYS_CONNECT: | |
1387c2c2 | 2487 | err = __sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2488 | break; |
2489 | case SYS_LISTEN: | |
2490 | err = sys_listen(a0, a1); | |
2491 | break; | |
2492 | case SYS_ACCEPT: | |
4541e805 DB |
2493 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2494 | (int __user *)a[2], 0); | |
89bddce5 SH |
2495 | break; |
2496 | case SYS_GETSOCKNAME: | |
2497 | err = | |
2498 | sys_getsockname(a0, (struct sockaddr __user *)a1, | |
2499 | (int __user *)a[2]); | |
2500 | break; | |
2501 | case SYS_GETPEERNAME: | |
2502 | err = | |
2503 | sys_getpeername(a0, (struct sockaddr __user *)a1, | |
2504 | (int __user *)a[2]); | |
2505 | break; | |
2506 | case SYS_SOCKETPAIR: | |
2507 | err = sys_socketpair(a0, a1, a[2], (int __user *)a[3]); | |
2508 | break; | |
2509 | case SYS_SEND: | |
2510 | err = sys_send(a0, (void __user *)a1, a[2], a[3]); | |
2511 | break; | |
2512 | case SYS_SENDTO: | |
211b634b DB |
2513 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2514 | (struct sockaddr __user *)a[4], a[5]); | |
89bddce5 SH |
2515 | break; |
2516 | case SYS_RECV: | |
2517 | err = sys_recv(a0, (void __user *)a1, a[2], a[3]); | |
2518 | break; | |
2519 | case SYS_RECVFROM: | |
7a09e1eb DB |
2520 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2521 | (struct sockaddr __user *)a[4], | |
2522 | (int __user *)a[5]); | |
89bddce5 SH |
2523 | break; |
2524 | case SYS_SHUTDOWN: | |
2525 | err = sys_shutdown(a0, a1); | |
2526 | break; | |
2527 | case SYS_SETSOCKOPT: | |
2528 | err = sys_setsockopt(a0, a1, a[2], (char __user *)a[3], a[4]); | |
2529 | break; | |
2530 | case SYS_GETSOCKOPT: | |
2531 | err = | |
2532 | sys_getsockopt(a0, a1, a[2], (char __user *)a[3], | |
2533 | (int __user *)a[4]); | |
2534 | break; | |
2535 | case SYS_SENDMSG: | |
666547ff | 2536 | err = sys_sendmsg(a0, (struct user_msghdr __user *)a1, a[2]); |
89bddce5 | 2537 | break; |
228e548e AB |
2538 | case SYS_SENDMMSG: |
2539 | err = sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3]); | |
2540 | break; | |
89bddce5 | 2541 | case SYS_RECVMSG: |
666547ff | 2542 | err = sys_recvmsg(a0, (struct user_msghdr __user *)a1, a[2]); |
89bddce5 | 2543 | break; |
a2e27255 ACM |
2544 | case SYS_RECVMMSG: |
2545 | err = sys_recvmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3], | |
2546 | (struct timespec __user *)a[4]); | |
2547 | break; | |
de11defe | 2548 | case SYS_ACCEPT4: |
4541e805 DB |
2549 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2550 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2551 | break; |
89bddce5 SH |
2552 | default: |
2553 | err = -EINVAL; | |
2554 | break; | |
1da177e4 LT |
2555 | } |
2556 | return err; | |
2557 | } | |
2558 | ||
89bddce5 | 2559 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2560 | |
55737fda SH |
2561 | /** |
2562 | * sock_register - add a socket protocol handler | |
2563 | * @ops: description of protocol | |
2564 | * | |
1da177e4 LT |
2565 | * This function is called by a protocol handler that wants to |
2566 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2567 | * socket interface. The value ops->family corresponds to the |
55737fda | 2568 | * socket system call protocol family. |
1da177e4 | 2569 | */ |
f0fd27d4 | 2570 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2571 | { |
2572 | int err; | |
2573 | ||
2574 | if (ops->family >= NPROTO) { | |
3410f22e | 2575 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2576 | return -ENOBUFS; |
2577 | } | |
55737fda SH |
2578 | |
2579 | spin_lock(&net_family_lock); | |
190683a9 ED |
2580 | if (rcu_dereference_protected(net_families[ops->family], |
2581 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2582 | err = -EEXIST; |
2583 | else { | |
cf778b00 | 2584 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2585 | err = 0; |
2586 | } | |
55737fda SH |
2587 | spin_unlock(&net_family_lock); |
2588 | ||
3410f22e | 2589 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2590 | return err; |
2591 | } | |
c6d409cf | 2592 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2593 | |
55737fda SH |
2594 | /** |
2595 | * sock_unregister - remove a protocol handler | |
2596 | * @family: protocol family to remove | |
2597 | * | |
1da177e4 LT |
2598 | * This function is called by a protocol handler that wants to |
2599 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2600 | * new socket creation. |
2601 | * | |
2602 | * If protocol handler is a module, then it can use module reference | |
2603 | * counts to protect against new references. If protocol handler is not | |
2604 | * a module then it needs to provide its own protection in | |
2605 | * the ops->create routine. | |
1da177e4 | 2606 | */ |
f0fd27d4 | 2607 | void sock_unregister(int family) |
1da177e4 | 2608 | { |
f0fd27d4 | 2609 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2610 | |
55737fda | 2611 | spin_lock(&net_family_lock); |
a9b3cd7f | 2612 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2613 | spin_unlock(&net_family_lock); |
2614 | ||
2615 | synchronize_rcu(); | |
2616 | ||
3410f22e | 2617 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2618 | } |
c6d409cf | 2619 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2620 | |
77d76ea3 | 2621 | static int __init sock_init(void) |
1da177e4 | 2622 | { |
b3e19d92 | 2623 | int err; |
2ca794e5 EB |
2624 | /* |
2625 | * Initialize the network sysctl infrastructure. | |
2626 | */ | |
2627 | err = net_sysctl_init(); | |
2628 | if (err) | |
2629 | goto out; | |
b3e19d92 | 2630 | |
1da177e4 | 2631 | /* |
89bddce5 | 2632 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2633 | */ |
2634 | skb_init(); | |
1da177e4 LT |
2635 | |
2636 | /* | |
89bddce5 | 2637 | * Initialize the protocols module. |
1da177e4 LT |
2638 | */ |
2639 | ||
2640 | init_inodecache(); | |
b3e19d92 NP |
2641 | |
2642 | err = register_filesystem(&sock_fs_type); | |
2643 | if (err) | |
2644 | goto out_fs; | |
1da177e4 | 2645 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2646 | if (IS_ERR(sock_mnt)) { |
2647 | err = PTR_ERR(sock_mnt); | |
2648 | goto out_mount; | |
2649 | } | |
77d76ea3 AK |
2650 | |
2651 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2652 | */ |
2653 | ||
2654 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
2655 | err = netfilter_init(); |
2656 | if (err) | |
2657 | goto out; | |
1da177e4 | 2658 | #endif |
cbeb321a | 2659 | |
408eccce | 2660 | ptp_classifier_init(); |
c1f19b51 | 2661 | |
b3e19d92 NP |
2662 | out: |
2663 | return err; | |
2664 | ||
2665 | out_mount: | |
2666 | unregister_filesystem(&sock_fs_type); | |
2667 | out_fs: | |
2668 | goto out; | |
1da177e4 LT |
2669 | } |
2670 | ||
77d76ea3 AK |
2671 | core_initcall(sock_init); /* early initcall */ |
2672 | ||
1da177e4 LT |
2673 | #ifdef CONFIG_PROC_FS |
2674 | void socket_seq_show(struct seq_file *seq) | |
2675 | { | |
648845ab TZ |
2676 | seq_printf(seq, "sockets: used %d\n", |
2677 | sock_inuse_get(seq->private)); | |
1da177e4 | 2678 | } |
89bddce5 | 2679 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 2680 | |
89bbfc95 | 2681 | #ifdef CONFIG_COMPAT |
6b96018b | 2682 | static int do_siocgstamp(struct net *net, struct socket *sock, |
644595f8 | 2683 | unsigned int cmd, void __user *up) |
7a229387 | 2684 | { |
7a229387 AB |
2685 | mm_segment_t old_fs = get_fs(); |
2686 | struct timeval ktv; | |
2687 | int err; | |
2688 | ||
2689 | set_fs(KERNEL_DS); | |
6b96018b | 2690 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&ktv); |
7a229387 | 2691 | set_fs(old_fs); |
644595f8 | 2692 | if (!err) |
ed6fe9d6 | 2693 | err = compat_put_timeval(&ktv, up); |
644595f8 | 2694 | |
7a229387 AB |
2695 | return err; |
2696 | } | |
2697 | ||
6b96018b | 2698 | static int do_siocgstampns(struct net *net, struct socket *sock, |
644595f8 | 2699 | unsigned int cmd, void __user *up) |
7a229387 | 2700 | { |
7a229387 AB |
2701 | mm_segment_t old_fs = get_fs(); |
2702 | struct timespec kts; | |
2703 | int err; | |
2704 | ||
2705 | set_fs(KERNEL_DS); | |
6b96018b | 2706 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&kts); |
7a229387 | 2707 | set_fs(old_fs); |
644595f8 | 2708 | if (!err) |
ed6fe9d6 | 2709 | err = compat_put_timespec(&kts, up); |
644595f8 | 2710 | |
7a229387 AB |
2711 | return err; |
2712 | } | |
2713 | ||
36fd633e | 2714 | static int compat_dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 2715 | { |
6b96018b | 2716 | struct compat_ifconf ifc32; |
7a229387 | 2717 | struct ifconf ifc; |
7a229387 AB |
2718 | int err; |
2719 | ||
6b96018b | 2720 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2721 | return -EFAULT; |
2722 | ||
36fd633e AV |
2723 | ifc.ifc_len = ifc32.ifc_len; |
2724 | ifc.ifc_req = compat_ptr(ifc32.ifcbuf); | |
7a229387 | 2725 | |
36fd633e AV |
2726 | rtnl_lock(); |
2727 | err = dev_ifconf(net, &ifc, sizeof(struct compat_ifreq)); | |
2728 | rtnl_unlock(); | |
7a229387 AB |
2729 | if (err) |
2730 | return err; | |
2731 | ||
36fd633e | 2732 | ifc32.ifc_len = ifc.ifc_len; |
6b96018b | 2733 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2734 | return -EFAULT; |
2735 | ||
2736 | return 0; | |
2737 | } | |
2738 | ||
6b96018b | 2739 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 2740 | { |
3a7da39d BH |
2741 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
2742 | bool convert_in = false, convert_out = false; | |
44c02a2c AV |
2743 | size_t buf_size = 0; |
2744 | struct ethtool_rxnfc __user *rxnfc = NULL; | |
2745 | struct ifreq ifr; | |
3a7da39d BH |
2746 | u32 rule_cnt = 0, actual_rule_cnt; |
2747 | u32 ethcmd; | |
7a229387 | 2748 | u32 data; |
3a7da39d | 2749 | int ret; |
7a229387 | 2750 | |
3a7da39d BH |
2751 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
2752 | return -EFAULT; | |
7a229387 | 2753 | |
3a7da39d BH |
2754 | compat_rxnfc = compat_ptr(data); |
2755 | ||
2756 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
2757 | return -EFAULT; |
2758 | ||
3a7da39d BH |
2759 | /* Most ethtool structures are defined without padding. |
2760 | * Unfortunately struct ethtool_rxnfc is an exception. | |
2761 | */ | |
2762 | switch (ethcmd) { | |
2763 | default: | |
2764 | break; | |
2765 | case ETHTOOL_GRXCLSRLALL: | |
2766 | /* Buffer size is variable */ | |
2767 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
2768 | return -EFAULT; | |
2769 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
2770 | return -ENOMEM; | |
2771 | buf_size += rule_cnt * sizeof(u32); | |
2772 | /* fall through */ | |
2773 | case ETHTOOL_GRXRINGS: | |
2774 | case ETHTOOL_GRXCLSRLCNT: | |
2775 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 2776 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
2777 | convert_out = true; |
2778 | /* fall through */ | |
2779 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
2780 | buf_size += sizeof(struct ethtool_rxnfc); |
2781 | convert_in = true; | |
44c02a2c | 2782 | rxnfc = compat_alloc_user_space(buf_size); |
3a7da39d BH |
2783 | break; |
2784 | } | |
2785 | ||
44c02a2c | 2786 | if (copy_from_user(&ifr.ifr_name, &ifr32->ifr_name, IFNAMSIZ)) |
7a229387 AB |
2787 | return -EFAULT; |
2788 | ||
44c02a2c | 2789 | ifr.ifr_data = convert_in ? rxnfc : (void __user *)compat_rxnfc; |
7a229387 | 2790 | |
3a7da39d | 2791 | if (convert_in) { |
127fe533 | 2792 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
2793 | * fs.ring_cookie and at the end of fs, but nowhere else. |
2794 | */ | |
127fe533 AD |
2795 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
2796 | sizeof(compat_rxnfc->fs.m_ext) != | |
2797 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
2798 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
2799 | BUILD_BUG_ON( |
2800 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
2801 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
2802 | offsetof(struct ethtool_rxnfc, fs.location) - | |
2803 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
2804 | ||
2805 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
2806 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
2807 | (void __user *)rxnfc) || | |
3a7da39d BH |
2808 | copy_in_user(&rxnfc->fs.ring_cookie, |
2809 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 SH |
2810 | (void __user *)(&rxnfc->fs.location + 1) - |
2811 | (void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2812 | copy_in_user(&rxnfc->rule_cnt, &compat_rxnfc->rule_cnt, |
2813 | sizeof(rxnfc->rule_cnt))) | |
2814 | return -EFAULT; | |
2815 | } | |
2816 | ||
44c02a2c | 2817 | ret = dev_ioctl(net, SIOCETHTOOL, &ifr, NULL); |
3a7da39d BH |
2818 | if (ret) |
2819 | return ret; | |
2820 | ||
2821 | if (convert_out) { | |
2822 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
2823 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
2824 | (const void __user *)rxnfc) || | |
3a7da39d BH |
2825 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
2826 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
2827 | (const void __user *)(&rxnfc->fs.location + 1) - |
2828 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2829 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
2830 | sizeof(rxnfc->rule_cnt))) | |
2831 | return -EFAULT; | |
2832 | ||
2833 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
2834 | /* As an optimisation, we only copy the actual | |
2835 | * number of rules that the underlying | |
2836 | * function returned. Since Mallory might | |
2837 | * change the rule count in user memory, we | |
2838 | * check that it is less than the rule count | |
2839 | * originally given (as the user buffer size), | |
2840 | * which has been range-checked. | |
2841 | */ | |
2842 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
2843 | return -EFAULT; | |
2844 | if (actual_rule_cnt < rule_cnt) | |
2845 | rule_cnt = actual_rule_cnt; | |
2846 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
2847 | &rxnfc->rule_locs[0], | |
2848 | rule_cnt * sizeof(u32))) | |
2849 | return -EFAULT; | |
2850 | } | |
2851 | } | |
2852 | ||
2853 | return 0; | |
7a229387 AB |
2854 | } |
2855 | ||
7a50a240 AB |
2856 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
2857 | { | |
7a50a240 | 2858 | compat_uptr_t uptr32; |
44c02a2c AV |
2859 | struct ifreq ifr; |
2860 | void __user *saved; | |
2861 | int err; | |
7a50a240 | 2862 | |
44c02a2c | 2863 | if (copy_from_user(&ifr, uifr32, sizeof(struct compat_ifreq))) |
7a50a240 AB |
2864 | return -EFAULT; |
2865 | ||
2866 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
2867 | return -EFAULT; | |
2868 | ||
44c02a2c AV |
2869 | saved = ifr.ifr_settings.ifs_ifsu.raw_hdlc; |
2870 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = compat_ptr(uptr32); | |
7a229387 | 2871 | |
44c02a2c AV |
2872 | err = dev_ioctl(net, SIOCWANDEV, &ifr, NULL); |
2873 | if (!err) { | |
2874 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = saved; | |
2875 | if (copy_to_user(uifr32, &ifr, sizeof(struct compat_ifreq))) | |
2876 | err = -EFAULT; | |
ccbd6a5a | 2877 | } |
44c02a2c | 2878 | return err; |
7a229387 AB |
2879 | } |
2880 | ||
590d4693 BH |
2881 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
2882 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 2883 | struct compat_ifreq __user *u_ifreq32) |
7a229387 | 2884 | { |
44c02a2c | 2885 | struct ifreq ifreq; |
7a229387 AB |
2886 | u32 data32; |
2887 | ||
44c02a2c | 2888 | if (copy_from_user(ifreq.ifr_name, u_ifreq32->ifr_name, IFNAMSIZ)) |
7a229387 | 2889 | return -EFAULT; |
44c02a2c | 2890 | if (get_user(data32, &u_ifreq32->ifr_data)) |
7a229387 | 2891 | return -EFAULT; |
44c02a2c | 2892 | ifreq.ifr_data = compat_ptr(data32); |
7a229387 | 2893 | |
44c02a2c | 2894 | return dev_ioctl(net, cmd, &ifreq, NULL); |
7a229387 AB |
2895 | } |
2896 | ||
a2116ed2 AB |
2897 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
2898 | struct compat_ifreq __user *uifr32) | |
2899 | { | |
2900 | struct ifreq ifr; | |
2901 | struct compat_ifmap __user *uifmap32; | |
a2116ed2 AB |
2902 | int err; |
2903 | ||
2904 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
2905 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
2906 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
2907 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
2908 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
2909 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
2910 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
2911 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
2912 | if (err) |
2913 | return -EFAULT; | |
2914 | ||
44c02a2c | 2915 | err = dev_ioctl(net, cmd, &ifr, NULL); |
a2116ed2 AB |
2916 | |
2917 | if (cmd == SIOCGIFMAP && !err) { | |
2918 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
2919 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
2920 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
2921 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
2922 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
2923 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
2924 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
2925 | if (err) |
2926 | err = -EFAULT; | |
2927 | } | |
2928 | return err; | |
2929 | } | |
2930 | ||
7a229387 | 2931 | struct rtentry32 { |
c6d409cf | 2932 | u32 rt_pad1; |
7a229387 AB |
2933 | struct sockaddr rt_dst; /* target address */ |
2934 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
2935 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
2936 | unsigned short rt_flags; |
2937 | short rt_pad2; | |
2938 | u32 rt_pad3; | |
2939 | unsigned char rt_tos; | |
2940 | unsigned char rt_class; | |
2941 | short rt_pad4; | |
2942 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 2943 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
2944 | u32 rt_mtu; /* per route MTU/Window */ |
2945 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
2946 | unsigned short rt_irtt; /* Initial RTT */ |
2947 | }; | |
2948 | ||
2949 | struct in6_rtmsg32 { | |
2950 | struct in6_addr rtmsg_dst; | |
2951 | struct in6_addr rtmsg_src; | |
2952 | struct in6_addr rtmsg_gateway; | |
2953 | u32 rtmsg_type; | |
2954 | u16 rtmsg_dst_len; | |
2955 | u16 rtmsg_src_len; | |
2956 | u32 rtmsg_metric; | |
2957 | u32 rtmsg_info; | |
2958 | u32 rtmsg_flags; | |
2959 | s32 rtmsg_ifindex; | |
2960 | }; | |
2961 | ||
6b96018b AB |
2962 | static int routing_ioctl(struct net *net, struct socket *sock, |
2963 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
2964 | { |
2965 | int ret; | |
2966 | void *r = NULL; | |
2967 | struct in6_rtmsg r6; | |
2968 | struct rtentry r4; | |
2969 | char devname[16]; | |
2970 | u32 rtdev; | |
2971 | mm_segment_t old_fs = get_fs(); | |
2972 | ||
6b96018b AB |
2973 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
2974 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 2975 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 2976 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
2977 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
2978 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
2979 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
2980 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
2981 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
2982 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
2983 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
2984 | |
2985 | r = (void *) &r6; | |
2986 | } else { /* ipv4 */ | |
6b96018b | 2987 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 2988 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 2989 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
2990 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
2991 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
2992 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
2993 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
2994 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
2995 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 2996 | if (rtdev) { |
c6d409cf | 2997 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 2998 | r4.rt_dev = (char __user __force *)devname; |
2999 | devname[15] = 0; | |
7a229387 AB |
3000 | } else |
3001 | r4.rt_dev = NULL; | |
3002 | ||
3003 | r = (void *) &r4; | |
3004 | } | |
3005 | ||
3006 | if (ret) { | |
3007 | ret = -EFAULT; | |
3008 | goto out; | |
3009 | } | |
3010 | ||
c6d409cf | 3011 | set_fs(KERNEL_DS); |
6b96018b | 3012 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3013 | set_fs(old_fs); |
7a229387 AB |
3014 | |
3015 | out: | |
7a229387 AB |
3016 | return ret; |
3017 | } | |
3018 | ||
3019 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3020 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3021 | * use compatible ioctls |
7a229387 | 3022 | */ |
6b96018b | 3023 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3024 | { |
6b96018b | 3025 | compat_ulong_t tmp; |
7a229387 | 3026 | |
6b96018b | 3027 | if (get_user(tmp, argp)) |
7a229387 AB |
3028 | return -EFAULT; |
3029 | if (tmp == BRCTL_GET_VERSION) | |
3030 | return BRCTL_VERSION + 1; | |
3031 | return -EINVAL; | |
3032 | } | |
3033 | ||
6b96018b AB |
3034 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3035 | unsigned int cmd, unsigned long arg) | |
3036 | { | |
3037 | void __user *argp = compat_ptr(arg); | |
3038 | struct sock *sk = sock->sk; | |
3039 | struct net *net = sock_net(sk); | |
7a229387 | 3040 | |
6b96018b | 3041 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3042 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3043 | |
3044 | switch (cmd) { | |
3045 | case SIOCSIFBR: | |
3046 | case SIOCGIFBR: | |
3047 | return old_bridge_ioctl(argp); | |
6b96018b | 3048 | case SIOCGIFCONF: |
36fd633e | 3049 | return compat_dev_ifconf(net, argp); |
6b96018b AB |
3050 | case SIOCETHTOOL: |
3051 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3052 | case SIOCWANDEV: |
3053 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3054 | case SIOCGIFMAP: |
3055 | case SIOCSIFMAP: | |
3056 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3057 | case SIOCADDRT: |
3058 | case SIOCDELRT: | |
3059 | return routing_ioctl(net, sock, cmd, argp); | |
3060 | case SIOCGSTAMP: | |
3061 | return do_siocgstamp(net, sock, cmd, argp); | |
3062 | case SIOCGSTAMPNS: | |
3063 | return do_siocgstampns(net, sock, cmd, argp); | |
590d4693 BH |
3064 | case SIOCBONDSLAVEINFOQUERY: |
3065 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3066 | case SIOCSHWTSTAMP: |
fd468c74 | 3067 | case SIOCGHWTSTAMP: |
590d4693 | 3068 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3069 | |
3070 | case FIOSETOWN: | |
3071 | case SIOCSPGRP: | |
3072 | case FIOGETOWN: | |
3073 | case SIOCGPGRP: | |
3074 | case SIOCBRADDBR: | |
3075 | case SIOCBRDELBR: | |
3076 | case SIOCGIFVLAN: | |
3077 | case SIOCSIFVLAN: | |
3078 | case SIOCADDDLCI: | |
3079 | case SIOCDELDLCI: | |
c62cce2c | 3080 | case SIOCGSKNS: |
6b96018b AB |
3081 | return sock_ioctl(file, cmd, arg); |
3082 | ||
3083 | case SIOCGIFFLAGS: | |
3084 | case SIOCSIFFLAGS: | |
3085 | case SIOCGIFMETRIC: | |
3086 | case SIOCSIFMETRIC: | |
3087 | case SIOCGIFMTU: | |
3088 | case SIOCSIFMTU: | |
3089 | case SIOCGIFMEM: | |
3090 | case SIOCSIFMEM: | |
3091 | case SIOCGIFHWADDR: | |
3092 | case SIOCSIFHWADDR: | |
3093 | case SIOCADDMULTI: | |
3094 | case SIOCDELMULTI: | |
3095 | case SIOCGIFINDEX: | |
6b96018b AB |
3096 | case SIOCGIFADDR: |
3097 | case SIOCSIFADDR: | |
3098 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3099 | case SIOCDIFADDR: |
6b96018b AB |
3100 | case SIOCGIFBRDADDR: |
3101 | case SIOCSIFBRDADDR: | |
3102 | case SIOCGIFDSTADDR: | |
3103 | case SIOCSIFDSTADDR: | |
3104 | case SIOCGIFNETMASK: | |
3105 | case SIOCSIFNETMASK: | |
3106 | case SIOCSIFPFLAGS: | |
3107 | case SIOCGIFPFLAGS: | |
3108 | case SIOCGIFTXQLEN: | |
3109 | case SIOCSIFTXQLEN: | |
3110 | case SIOCBRADDIF: | |
3111 | case SIOCBRDELIF: | |
9177efd3 AB |
3112 | case SIOCSIFNAME: |
3113 | case SIOCGMIIPHY: | |
3114 | case SIOCGMIIREG: | |
3115 | case SIOCSMIIREG: | |
6b96018b AB |
3116 | case SIOCSARP: |
3117 | case SIOCGARP: | |
3118 | case SIOCDARP: | |
6b96018b | 3119 | case SIOCATMARK: |
f92d4fc9 AV |
3120 | case SIOCBONDENSLAVE: |
3121 | case SIOCBONDRELEASE: | |
3122 | case SIOCBONDSETHWADDR: | |
3123 | case SIOCBONDCHANGEACTIVE: | |
4cf808e7 | 3124 | case SIOCGIFNAME: |
9177efd3 AB |
3125 | return sock_do_ioctl(net, sock, cmd, arg); |
3126 | } | |
3127 | ||
6b96018b AB |
3128 | return -ENOIOCTLCMD; |
3129 | } | |
7a229387 | 3130 | |
95c96174 | 3131 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3132 | unsigned long arg) |
89bbfc95 SP |
3133 | { |
3134 | struct socket *sock = file->private_data; | |
3135 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3136 | struct sock *sk; |
3137 | struct net *net; | |
3138 | ||
3139 | sk = sock->sk; | |
3140 | net = sock_net(sk); | |
89bbfc95 SP |
3141 | |
3142 | if (sock->ops->compat_ioctl) | |
3143 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3144 | ||
87de87d5 DM |
3145 | if (ret == -ENOIOCTLCMD && |
3146 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3147 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3148 | ||
6b96018b AB |
3149 | if (ret == -ENOIOCTLCMD) |
3150 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3151 | ||
89bbfc95 SP |
3152 | return ret; |
3153 | } | |
3154 | #endif | |
3155 | ||
ac5a488e SS |
3156 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3157 | { | |
3158 | return sock->ops->bind(sock, addr, addrlen); | |
3159 | } | |
c6d409cf | 3160 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e SS |
3161 | |
3162 | int kernel_listen(struct socket *sock, int backlog) | |
3163 | { | |
3164 | return sock->ops->listen(sock, backlog); | |
3165 | } | |
c6d409cf | 3166 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e SS |
3167 | |
3168 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) | |
3169 | { | |
3170 | struct sock *sk = sock->sk; | |
3171 | int err; | |
3172 | ||
3173 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3174 | newsock); | |
3175 | if (err < 0) | |
3176 | goto done; | |
3177 | ||
cdfbabfb | 3178 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3179 | if (err < 0) { |
3180 | sock_release(*newsock); | |
fa8705b0 | 3181 | *newsock = NULL; |
ac5a488e SS |
3182 | goto done; |
3183 | } | |
3184 | ||
3185 | (*newsock)->ops = sock->ops; | |
1b08534e | 3186 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3187 | |
3188 | done: | |
3189 | return err; | |
3190 | } | |
c6d409cf | 3191 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e SS |
3192 | |
3193 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, | |
4768fbcb | 3194 | int flags) |
ac5a488e SS |
3195 | { |
3196 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3197 | } | |
c6d409cf | 3198 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e SS |
3199 | |
3200 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr, | |
3201 | int *addrlen) | |
3202 | { | |
3203 | return sock->ops->getname(sock, addr, addrlen, 0); | |
3204 | } | |
c6d409cf | 3205 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e SS |
3206 | |
3207 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr, | |
3208 | int *addrlen) | |
3209 | { | |
3210 | return sock->ops->getname(sock, addr, addrlen, 1); | |
3211 | } | |
c6d409cf | 3212 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e SS |
3213 | |
3214 | int kernel_getsockopt(struct socket *sock, int level, int optname, | |
3215 | char *optval, int *optlen) | |
3216 | { | |
3217 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3218 | char __user *uoptval; |
3219 | int __user *uoptlen; | |
ac5a488e SS |
3220 | int err; |
3221 | ||
fb8621bb NK |
3222 | uoptval = (char __user __force *) optval; |
3223 | uoptlen = (int __user __force *) optlen; | |
3224 | ||
ac5a488e SS |
3225 | set_fs(KERNEL_DS); |
3226 | if (level == SOL_SOCKET) | |
fb8621bb | 3227 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3228 | else |
fb8621bb NK |
3229 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3230 | uoptlen); | |
ac5a488e SS |
3231 | set_fs(oldfs); |
3232 | return err; | |
3233 | } | |
c6d409cf | 3234 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e SS |
3235 | |
3236 | int kernel_setsockopt(struct socket *sock, int level, int optname, | |
b7058842 | 3237 | char *optval, unsigned int optlen) |
ac5a488e SS |
3238 | { |
3239 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3240 | char __user *uoptval; |
ac5a488e SS |
3241 | int err; |
3242 | ||
fb8621bb NK |
3243 | uoptval = (char __user __force *) optval; |
3244 | ||
ac5a488e SS |
3245 | set_fs(KERNEL_DS); |
3246 | if (level == SOL_SOCKET) | |
fb8621bb | 3247 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3248 | else |
fb8621bb | 3249 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3250 | optlen); |
3251 | set_fs(oldfs); | |
3252 | return err; | |
3253 | } | |
c6d409cf | 3254 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e SS |
3255 | |
3256 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, | |
3257 | size_t size, int flags) | |
3258 | { | |
3259 | if (sock->ops->sendpage) | |
3260 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3261 | ||
3262 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3263 | } | |
c6d409cf | 3264 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3265 | |
306b13eb TH |
3266 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3267 | size_t size, int flags) | |
3268 | { | |
3269 | struct socket *sock = sk->sk_socket; | |
3270 | ||
3271 | if (sock->ops->sendpage_locked) | |
3272 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3273 | flags); | |
3274 | ||
3275 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3276 | } | |
3277 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3278 | ||
91cf45f0 TM |
3279 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3280 | { | |
3281 | return sock->ops->shutdown(sock, how); | |
3282 | } | |
91cf45f0 | 3283 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 P |
3284 | |
3285 | /* This routine returns the IP overhead imposed by a socket i.e. | |
3286 | * the length of the underlying IP header, depending on whether | |
3287 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
57240d00 | 3288 | * on at the socket. Assumes that the caller has a lock on the socket. |
113c3075 P |
3289 | */ |
3290 | u32 kernel_sock_ip_overhead(struct sock *sk) | |
3291 | { | |
3292 | struct inet_sock *inet; | |
3293 | struct ip_options_rcu *opt; | |
3294 | u32 overhead = 0; | |
113c3075 P |
3295 | #if IS_ENABLED(CONFIG_IPV6) |
3296 | struct ipv6_pinfo *np; | |
3297 | struct ipv6_txoptions *optv6 = NULL; | |
3298 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3299 | ||
3300 | if (!sk) | |
3301 | return overhead; | |
3302 | ||
113c3075 P |
3303 | switch (sk->sk_family) { |
3304 | case AF_INET: | |
3305 | inet = inet_sk(sk); | |
3306 | overhead += sizeof(struct iphdr); | |
3307 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3308 | sock_owned_by_user(sk)); |
113c3075 P |
3309 | if (opt) |
3310 | overhead += opt->opt.optlen; | |
3311 | return overhead; | |
3312 | #if IS_ENABLED(CONFIG_IPV6) | |
3313 | case AF_INET6: | |
3314 | np = inet6_sk(sk); | |
3315 | overhead += sizeof(struct ipv6hdr); | |
3316 | if (np) | |
3317 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3318 | sock_owned_by_user(sk)); |
113c3075 P |
3319 | if (optv6) |
3320 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3321 | return overhead; | |
3322 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3323 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3324 | return overhead; | |
3325 | } | |
3326 | } | |
3327 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |