]>
Commit | Line | Data |
---|---|---|
2874c5fd | 1 | // SPDX-License-Identifier: GPL-2.0-or-later |
1da177e4 LT |
2 | /* |
3 | * NET An implementation of the SOCKET network access protocol. | |
4 | * | |
5 | * Version: @(#)socket.c 1.1.93 18/02/95 | |
6 | * | |
7 | * Authors: Orest Zborowski, <obz@Kodak.COM> | |
02c30a84 | 8 | * Ross Biro |
1da177e4 LT |
9 | * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> |
10 | * | |
11 | * Fixes: | |
12 | * Anonymous : NOTSOCK/BADF cleanup. Error fix in | |
13 | * shutdown() | |
14 | * Alan Cox : verify_area() fixes | |
15 | * Alan Cox : Removed DDI | |
16 | * Jonathan Kamens : SOCK_DGRAM reconnect bug | |
17 | * Alan Cox : Moved a load of checks to the very | |
18 | * top level. | |
19 | * Alan Cox : Move address structures to/from user | |
20 | * mode above the protocol layers. | |
21 | * Rob Janssen : Allow 0 length sends. | |
22 | * Alan Cox : Asynchronous I/O support (cribbed from the | |
23 | * tty drivers). | |
24 | * Niibe Yutaka : Asynchronous I/O for writes (4.4BSD style) | |
25 | * Jeff Uphoff : Made max number of sockets command-line | |
26 | * configurable. | |
27 | * Matti Aarnio : Made the number of sockets dynamic, | |
28 | * to be allocated when needed, and mr. | |
29 | * Uphoff's max is used as max to be | |
30 | * allowed to allocate. | |
31 | * Linus : Argh. removed all the socket allocation | |
32 | * altogether: it's in the inode now. | |
33 | * Alan Cox : Made sock_alloc()/sock_release() public | |
34 | * for NetROM and future kernel nfsd type | |
35 | * stuff. | |
36 | * Alan Cox : sendmsg/recvmsg basics. | |
37 | * Tom Dyas : Export net symbols. | |
38 | * Marcin Dalecki : Fixed problems with CONFIG_NET="n". | |
39 | * Alan Cox : Added thread locking to sys_* calls | |
40 | * for sockets. May have errors at the | |
41 | * moment. | |
42 | * Kevin Buhr : Fixed the dumb errors in the above. | |
43 | * Andi Kleen : Some small cleanups, optimizations, | |
44 | * and fixed a copy_from_user() bug. | |
45 | * Tigran Aivazian : sys_send(args) calls sys_sendto(args, NULL, 0) | |
89bddce5 | 46 | * Tigran Aivazian : Made listen(2) backlog sanity checks |
1da177e4 LT |
47 | * protocol-independent |
48 | * | |
1da177e4 | 49 | * This module is effectively the top level interface to the BSD socket |
89bddce5 | 50 | * paradigm. |
1da177e4 LT |
51 | * |
52 | * Based upon Swansea University Computer Society NET3.039 | |
53 | */ | |
54 | ||
1da177e4 | 55 | #include <linux/mm.h> |
1da177e4 LT |
56 | #include <linux/socket.h> |
57 | #include <linux/file.h> | |
58 | #include <linux/net.h> | |
59 | #include <linux/interrupt.h> | |
aaca0bdc | 60 | #include <linux/thread_info.h> |
55737fda | 61 | #include <linux/rcupdate.h> |
1da177e4 LT |
62 | #include <linux/netdevice.h> |
63 | #include <linux/proc_fs.h> | |
64 | #include <linux/seq_file.h> | |
4a3e2f71 | 65 | #include <linux/mutex.h> |
1da177e4 | 66 | #include <linux/if_bridge.h> |
20380731 ACM |
67 | #include <linux/if_frad.h> |
68 | #include <linux/if_vlan.h> | |
408eccce | 69 | #include <linux/ptp_classify.h> |
1da177e4 LT |
70 | #include <linux/init.h> |
71 | #include <linux/poll.h> | |
72 | #include <linux/cache.h> | |
73 | #include <linux/module.h> | |
74 | #include <linux/highmem.h> | |
1da177e4 LT |
75 | #include <linux/mount.h> |
76 | #include <linux/security.h> | |
77 | #include <linux/syscalls.h> | |
78 | #include <linux/compat.h> | |
79 | #include <linux/kmod.h> | |
3ec3b2fb | 80 | #include <linux/audit.h> |
d86b5e0e | 81 | #include <linux/wireless.h> |
1b8d7ae4 | 82 | #include <linux/nsproxy.h> |
1fd7317d | 83 | #include <linux/magic.h> |
5a0e3ad6 | 84 | #include <linux/slab.h> |
600e1779 | 85 | #include <linux/xattr.h> |
c8e8cd57 | 86 | #include <linux/nospec.h> |
8c3c447b | 87 | #include <linux/indirect_call_wrapper.h> |
1da177e4 | 88 | |
7c0f6ba6 | 89 | #include <linux/uaccess.h> |
1da177e4 LT |
90 | #include <asm/unistd.h> |
91 | ||
92 | #include <net/compat.h> | |
87de87d5 | 93 | #include <net/wext.h> |
f8451725 | 94 | #include <net/cls_cgroup.h> |
1da177e4 LT |
95 | |
96 | #include <net/sock.h> | |
97 | #include <linux/netfilter.h> | |
98 | ||
6b96018b AB |
99 | #include <linux/if_tun.h> |
100 | #include <linux/ipv6_route.h> | |
101 | #include <linux/route.h> | |
6b96018b | 102 | #include <linux/sockios.h> |
076bb0c8 | 103 | #include <net/busy_poll.h> |
f24b9be5 | 104 | #include <linux/errqueue.h> |
06021292 | 105 | |
8c3c447b PA |
106 | /* proto_ops for ipv4 and ipv6 use the same {recv,send}msg function */ |
107 | #if IS_ENABLED(CONFIG_INET) | |
108 | #define INDIRECT_CALL_INET4(f, f1, ...) INDIRECT_CALL_1(f, f1, __VA_ARGS__) | |
109 | #else | |
110 | #define INDIRECT_CALL_INET4(f, f1, ...) f(__VA_ARGS__) | |
111 | #endif | |
112 | ||
e0d1095a | 113 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
114 | unsigned int sysctl_net_busy_read __read_mostly; |
115 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 116 | #endif |
6b96018b | 117 | |
8ae5e030 AV |
118 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
119 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 120 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
121 | |
122 | static int sock_close(struct inode *inode, struct file *file); | |
a11e1d43 LT |
123 | static __poll_t sock_poll(struct file *file, |
124 | struct poll_table_struct *wait); | |
89bddce5 | 125 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
126 | #ifdef CONFIG_COMPAT |
127 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 128 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 129 | #endif |
1da177e4 | 130 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
131 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
132 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 133 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 134 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 135 | unsigned int flags); |
1da177e4 | 136 | |
1da177e4 LT |
137 | /* |
138 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
139 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
140 | */ | |
141 | ||
da7071d7 | 142 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
143 | .owner = THIS_MODULE, |
144 | .llseek = no_llseek, | |
8ae5e030 AV |
145 | .read_iter = sock_read_iter, |
146 | .write_iter = sock_write_iter, | |
1da177e4 LT |
147 | .poll = sock_poll, |
148 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
149 | #ifdef CONFIG_COMPAT |
150 | .compat_ioctl = compat_sock_ioctl, | |
151 | #endif | |
1da177e4 | 152 | .mmap = sock_mmap, |
1da177e4 LT |
153 | .release = sock_close, |
154 | .fasync = sock_fasync, | |
5274f052 JA |
155 | .sendpage = sock_sendpage, |
156 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 157 | .splice_read = sock_splice_read, |
1da177e4 LT |
158 | }; |
159 | ||
160 | /* | |
161 | * The protocol list. Each protocol is registered in here. | |
162 | */ | |
163 | ||
1da177e4 | 164 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 165 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 166 | |
1da177e4 | 167 | /* |
89bddce5 SH |
168 | * Support routines. |
169 | * Move socket addresses back and forth across the kernel/user | |
170 | * divide and look after the messy bits. | |
1da177e4 LT |
171 | */ |
172 | ||
1da177e4 LT |
173 | /** |
174 | * move_addr_to_kernel - copy a socket address into kernel space | |
175 | * @uaddr: Address in user space | |
176 | * @kaddr: Address in kernel space | |
177 | * @ulen: Length in user space | |
178 | * | |
179 | * The address is copied into kernel space. If the provided address is | |
180 | * too long an error code of -EINVAL is returned. If the copy gives | |
181 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
182 | */ | |
183 | ||
43db362d | 184 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 185 | { |
230b1839 | 186 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 187 | return -EINVAL; |
89bddce5 | 188 | if (ulen == 0) |
1da177e4 | 189 | return 0; |
89bddce5 | 190 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 191 | return -EFAULT; |
3ec3b2fb | 192 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
193 | } |
194 | ||
195 | /** | |
196 | * move_addr_to_user - copy an address to user space | |
197 | * @kaddr: kernel space address | |
198 | * @klen: length of address in kernel | |
199 | * @uaddr: user space address | |
200 | * @ulen: pointer to user length field | |
201 | * | |
202 | * The value pointed to by ulen on entry is the buffer length available. | |
203 | * This is overwritten with the buffer space used. -EINVAL is returned | |
204 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
205 | * is returned if either the buffer or the length field are not | |
206 | * accessible. | |
207 | * After copying the data up to the limit the user specifies, the true | |
208 | * length of the data is written over the length limit the user | |
209 | * specified. Zero is returned for a success. | |
210 | */ | |
89bddce5 | 211 | |
43db362d | 212 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 213 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
214 | { |
215 | int err; | |
216 | int len; | |
217 | ||
68c6beb3 | 218 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
219 | err = get_user(len, ulen); |
220 | if (err) | |
1da177e4 | 221 | return err; |
89bddce5 SH |
222 | if (len > klen) |
223 | len = klen; | |
68c6beb3 | 224 | if (len < 0) |
1da177e4 | 225 | return -EINVAL; |
89bddce5 | 226 | if (len) { |
d6fe3945 SG |
227 | if (audit_sockaddr(klen, kaddr)) |
228 | return -ENOMEM; | |
89bddce5 | 229 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
230 | return -EFAULT; |
231 | } | |
232 | /* | |
89bddce5 SH |
233 | * "fromlen shall refer to the value before truncation.." |
234 | * 1003.1g | |
1da177e4 LT |
235 | */ |
236 | return __put_user(klen, ulen); | |
237 | } | |
238 | ||
08009a76 | 239 | static struct kmem_cache *sock_inode_cachep __ro_after_init; |
1da177e4 LT |
240 | |
241 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
242 | { | |
243 | struct socket_alloc *ei; | |
eaefd110 | 244 | struct socket_wq *wq; |
89bddce5 | 245 | |
e94b1766 | 246 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
247 | if (!ei) |
248 | return NULL; | |
eaefd110 ED |
249 | wq = kmalloc(sizeof(*wq), GFP_KERNEL); |
250 | if (!wq) { | |
43815482 ED |
251 | kmem_cache_free(sock_inode_cachep, ei); |
252 | return NULL; | |
253 | } | |
eaefd110 ED |
254 | init_waitqueue_head(&wq->wait); |
255 | wq->fasync_list = NULL; | |
574aab1e | 256 | wq->flags = 0; |
e6476c21 | 257 | ei->socket.wq = wq; |
89bddce5 | 258 | |
1da177e4 LT |
259 | ei->socket.state = SS_UNCONNECTED; |
260 | ei->socket.flags = 0; | |
261 | ei->socket.ops = NULL; | |
262 | ei->socket.sk = NULL; | |
263 | ei->socket.file = NULL; | |
1da177e4 LT |
264 | |
265 | return &ei->vfs_inode; | |
266 | } | |
267 | ||
268 | static void sock_destroy_inode(struct inode *inode) | |
269 | { | |
43815482 ED |
270 | struct socket_alloc *ei; |
271 | ||
272 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
e6476c21 | 273 | kfree_rcu(ei->socket.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 | ||
8a3c245c PT |
389 | /** |
390 | * sock_alloc_file - Bind a &socket to a &file | |
391 | * @sock: socket | |
392 | * @flags: file status flags | |
393 | * @dname: protocol name | |
394 | * | |
395 | * Returns the &file bound with @sock, implicitly storing it | |
396 | * in sock->file. If dname is %NULL, sets to "". | |
397 | * On failure the return is a ERR pointer (see linux/err.h). | |
398 | * This function uses GFP_KERNEL internally. | |
399 | */ | |
400 | ||
aab174f0 | 401 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 402 | { |
7cbe66b6 | 403 | struct file *file; |
1da177e4 | 404 | |
d93aa9d8 AV |
405 | if (!dname) |
406 | dname = sock->sk ? sock->sk->sk_prot_creator->name : ""; | |
39d8c1b6 | 407 | |
d93aa9d8 AV |
408 | file = alloc_file_pseudo(SOCK_INODE(sock), sock_mnt, dname, |
409 | O_RDWR | (flags & O_NONBLOCK), | |
410 | &socket_file_ops); | |
b5ffe634 | 411 | if (IS_ERR(file)) { |
8e1611e2 | 412 | sock_release(sock); |
39b65252 | 413 | return file; |
cc3808f8 AV |
414 | } |
415 | ||
416 | sock->file = file; | |
39d8c1b6 | 417 | file->private_data = sock; |
28407630 | 418 | return file; |
39d8c1b6 | 419 | } |
56b31d1c | 420 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 421 | |
56b31d1c | 422 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
423 | { |
424 | struct file *newfile; | |
28407630 | 425 | int fd = get_unused_fd_flags(flags); |
ce4bb04c AV |
426 | if (unlikely(fd < 0)) { |
427 | sock_release(sock); | |
28407630 | 428 | return fd; |
ce4bb04c | 429 | } |
39d8c1b6 | 430 | |
aab174f0 | 431 | newfile = sock_alloc_file(sock, flags, NULL); |
4546e44c | 432 | if (!IS_ERR(newfile)) { |
39d8c1b6 | 433 | fd_install(fd, newfile); |
28407630 AV |
434 | return fd; |
435 | } | |
7cbe66b6 | 436 | |
28407630 AV |
437 | put_unused_fd(fd); |
438 | return PTR_ERR(newfile); | |
1da177e4 LT |
439 | } |
440 | ||
8a3c245c PT |
441 | /** |
442 | * sock_from_file - Return the &socket bounded to @file. | |
443 | * @file: file | |
444 | * @err: pointer to an error code return | |
445 | * | |
446 | * On failure returns %NULL and assigns -ENOTSOCK to @err. | |
447 | */ | |
448 | ||
406a3c63 | 449 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 450 | { |
6cb153ca BL |
451 | if (file->f_op == &socket_file_ops) |
452 | return file->private_data; /* set in sock_map_fd */ | |
453 | ||
23bb80d2 ED |
454 | *err = -ENOTSOCK; |
455 | return NULL; | |
6cb153ca | 456 | } |
406a3c63 | 457 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 458 | |
1da177e4 | 459 | /** |
c6d409cf | 460 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
461 | * @fd: file handle |
462 | * @err: pointer to an error code return | |
463 | * | |
464 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 465 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
466 | * with a negative errno code and NULL is returned. The function checks |
467 | * for both invalid handles and passing a handle which is not a socket. | |
468 | * | |
469 | * On a success the socket object pointer is returned. | |
470 | */ | |
471 | ||
472 | struct socket *sockfd_lookup(int fd, int *err) | |
473 | { | |
474 | struct file *file; | |
1da177e4 LT |
475 | struct socket *sock; |
476 | ||
89bddce5 SH |
477 | file = fget(fd); |
478 | if (!file) { | |
1da177e4 LT |
479 | *err = -EBADF; |
480 | return NULL; | |
481 | } | |
89bddce5 | 482 | |
6cb153ca BL |
483 | sock = sock_from_file(file, err); |
484 | if (!sock) | |
1da177e4 | 485 | fput(file); |
6cb153ca BL |
486 | return sock; |
487 | } | |
c6d409cf | 488 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 489 | |
6cb153ca BL |
490 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
491 | { | |
00e188ef | 492 | struct fd f = fdget(fd); |
6cb153ca BL |
493 | struct socket *sock; |
494 | ||
3672558c | 495 | *err = -EBADF; |
00e188ef AV |
496 | if (f.file) { |
497 | sock = sock_from_file(f.file, err); | |
498 | if (likely(sock)) { | |
499 | *fput_needed = f.flags; | |
6cb153ca | 500 | return sock; |
00e188ef AV |
501 | } |
502 | fdput(f); | |
1da177e4 | 503 | } |
6cb153ca | 504 | return NULL; |
1da177e4 LT |
505 | } |
506 | ||
600e1779 MY |
507 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
508 | size_t size) | |
509 | { | |
510 | ssize_t len; | |
511 | ssize_t used = 0; | |
512 | ||
c5ef6035 | 513 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
514 | if (len < 0) |
515 | return len; | |
516 | used += len; | |
517 | if (buffer) { | |
518 | if (size < used) | |
519 | return -ERANGE; | |
520 | buffer += len; | |
521 | } | |
522 | ||
523 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
524 | used += len; | |
525 | if (buffer) { | |
526 | if (size < used) | |
527 | return -ERANGE; | |
528 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
529 | buffer += len; | |
530 | } | |
531 | ||
532 | return used; | |
533 | } | |
534 | ||
dc647ec8 | 535 | static int sockfs_setattr(struct dentry *dentry, struct iattr *iattr) |
86741ec2 LC |
536 | { |
537 | int err = simple_setattr(dentry, iattr); | |
538 | ||
e1a3a60a | 539 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
540 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
541 | ||
6d8c50dc CW |
542 | if (sock->sk) |
543 | sock->sk->sk_uid = iattr->ia_uid; | |
544 | else | |
545 | err = -ENOENT; | |
86741ec2 LC |
546 | } |
547 | ||
548 | return err; | |
549 | } | |
550 | ||
600e1779 | 551 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 552 | .listxattr = sockfs_listxattr, |
86741ec2 | 553 | .setattr = sockfs_setattr, |
600e1779 MY |
554 | }; |
555 | ||
1da177e4 | 556 | /** |
8a3c245c | 557 | * sock_alloc - allocate a socket |
89bddce5 | 558 | * |
1da177e4 LT |
559 | * Allocate a new inode and socket object. The two are bound together |
560 | * and initialised. The socket is then returned. If we are out of inodes | |
8a3c245c | 561 | * NULL is returned. This functions uses GFP_KERNEL internally. |
1da177e4 LT |
562 | */ |
563 | ||
f4a00aac | 564 | struct socket *sock_alloc(void) |
1da177e4 | 565 | { |
89bddce5 SH |
566 | struct inode *inode; |
567 | struct socket *sock; | |
1da177e4 | 568 | |
a209dfc7 | 569 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
570 | if (!inode) |
571 | return NULL; | |
572 | ||
573 | sock = SOCKET_I(inode); | |
574 | ||
85fe4025 | 575 | inode->i_ino = get_next_ino(); |
89bddce5 | 576 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
577 | inode->i_uid = current_fsuid(); |
578 | inode->i_gid = current_fsgid(); | |
600e1779 | 579 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 580 | |
1da177e4 LT |
581 | return sock; |
582 | } | |
f4a00aac | 583 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 584 | |
1da177e4 | 585 | /** |
8a3c245c | 586 | * sock_release - close a socket |
1da177e4 LT |
587 | * @sock: socket to close |
588 | * | |
589 | * The socket is released from the protocol stack if it has a release | |
590 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 591 | * an inode not a file. |
1da177e4 | 592 | */ |
89bddce5 | 593 | |
6d8c50dc | 594 | static void __sock_release(struct socket *sock, struct inode *inode) |
1da177e4 LT |
595 | { |
596 | if (sock->ops) { | |
597 | struct module *owner = sock->ops->owner; | |
598 | ||
6d8c50dc CW |
599 | if (inode) |
600 | inode_lock(inode); | |
1da177e4 | 601 | sock->ops->release(sock); |
ff7b11aa | 602 | sock->sk = NULL; |
6d8c50dc CW |
603 | if (inode) |
604 | inode_unlock(inode); | |
1da177e4 LT |
605 | sock->ops = NULL; |
606 | module_put(owner); | |
607 | } | |
608 | ||
e6476c21 | 609 | if (sock->wq->fasync_list) |
3410f22e | 610 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 611 | |
1da177e4 LT |
612 | if (!sock->file) { |
613 | iput(SOCK_INODE(sock)); | |
614 | return; | |
615 | } | |
89bddce5 | 616 | sock->file = NULL; |
1da177e4 | 617 | } |
6d8c50dc CW |
618 | |
619 | void sock_release(struct socket *sock) | |
620 | { | |
621 | __sock_release(sock, NULL); | |
622 | } | |
c6d409cf | 623 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 624 | |
c14ac945 | 625 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 626 | { |
140c55d4 ED |
627 | u8 flags = *tx_flags; |
628 | ||
c14ac945 | 629 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
630 | flags |= SKBTX_HW_TSTAMP; |
631 | ||
c14ac945 | 632 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
633 | flags |= SKBTX_SW_TSTAMP; |
634 | ||
c14ac945 | 635 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
636 | flags |= SKBTX_SCHED_TSTAMP; |
637 | ||
140c55d4 | 638 | *tx_flags = flags; |
20d49473 | 639 | } |
67cc0d40 | 640 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 641 | |
8c3c447b PA |
642 | INDIRECT_CALLABLE_DECLARE(int inet_sendmsg(struct socket *, struct msghdr *, |
643 | size_t)); | |
d8725c86 | 644 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 645 | { |
8c3c447b PA |
646 | int ret = INDIRECT_CALL_INET4(sock->ops->sendmsg, inet_sendmsg, sock, |
647 | msg, msg_data_left(msg)); | |
d8725c86 AV |
648 | BUG_ON(ret == -EIOCBQUEUED); |
649 | return ret; | |
1da177e4 LT |
650 | } |
651 | ||
85806af0 RD |
652 | /** |
653 | * sock_sendmsg - send a message through @sock | |
654 | * @sock: socket | |
655 | * @msg: message to send | |
656 | * | |
657 | * Sends @msg through @sock, passing through LSM. | |
658 | * Returns the number of bytes sent, or an error code. | |
659 | */ | |
d8725c86 | 660 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 661 | { |
d8725c86 | 662 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 663 | msg_data_left(msg)); |
228e548e | 664 | |
d8725c86 | 665 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 666 | } |
c6d409cf | 667 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 | 668 | |
8a3c245c PT |
669 | /** |
670 | * kernel_sendmsg - send a message through @sock (kernel-space) | |
671 | * @sock: socket | |
672 | * @msg: message header | |
673 | * @vec: kernel vec | |
674 | * @num: vec array length | |
675 | * @size: total message data size | |
676 | * | |
677 | * Builds the message data with @vec and sends it through @sock. | |
678 | * Returns the number of bytes sent, or an error code. | |
679 | */ | |
680 | ||
1da177e4 LT |
681 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
682 | struct kvec *vec, size_t num, size_t size) | |
683 | { | |
aa563d7b | 684 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
d8725c86 | 685 | return sock_sendmsg(sock, msg); |
1da177e4 | 686 | } |
c6d409cf | 687 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 688 | |
8a3c245c PT |
689 | /** |
690 | * kernel_sendmsg_locked - send a message through @sock (kernel-space) | |
691 | * @sk: sock | |
692 | * @msg: message header | |
693 | * @vec: output s/g array | |
694 | * @num: output s/g array length | |
695 | * @size: total message data size | |
696 | * | |
697 | * Builds the message data with @vec and sends it through @sock. | |
698 | * Returns the number of bytes sent, or an error code. | |
699 | * Caller must hold @sk. | |
700 | */ | |
701 | ||
306b13eb TH |
702 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
703 | struct kvec *vec, size_t num, size_t size) | |
704 | { | |
705 | struct socket *sock = sk->sk_socket; | |
706 | ||
707 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 708 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb | 709 | |
aa563d7b | 710 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
306b13eb TH |
711 | |
712 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
713 | } | |
714 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
715 | ||
8605330a SHY |
716 | static bool skb_is_err_queue(const struct sk_buff *skb) |
717 | { | |
718 | /* pkt_type of skbs enqueued on the error queue are set to | |
719 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
720 | * in recvmsg, since skbs received on a local socket will never | |
721 | * have a pkt_type of PACKET_OUTGOING. | |
722 | */ | |
723 | return skb->pkt_type == PACKET_OUTGOING; | |
724 | } | |
725 | ||
b50a5c70 ML |
726 | /* On transmit, software and hardware timestamps are returned independently. |
727 | * As the two skb clones share the hardware timestamp, which may be updated | |
728 | * before the software timestamp is received, a hardware TX timestamp may be | |
729 | * returned only if there is no software TX timestamp. Ignore false software | |
730 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
7f1bc6e9 | 731 | * option SO_TIMESTAMP_OLD(NS) is enabled on the socket, even when the skb has a |
b50a5c70 ML |
732 | * hardware timestamp. |
733 | */ | |
734 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
735 | { | |
736 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
737 | } | |
738 | ||
aad9c8c4 ML |
739 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb) |
740 | { | |
741 | struct scm_ts_pktinfo ts_pktinfo; | |
742 | struct net_device *orig_dev; | |
743 | ||
744 | if (!skb_mac_header_was_set(skb)) | |
745 | return; | |
746 | ||
747 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
748 | ||
749 | rcu_read_lock(); | |
750 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
751 | if (orig_dev) | |
752 | ts_pktinfo.if_index = orig_dev->ifindex; | |
753 | rcu_read_unlock(); | |
754 | ||
755 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
756 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
757 | sizeof(ts_pktinfo), &ts_pktinfo); | |
758 | } | |
759 | ||
92f37fd2 ED |
760 | /* |
761 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
762 | */ | |
763 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
764 | struct sk_buff *skb) | |
765 | { | |
20d49473 | 766 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
887feae3 | 767 | int new_tstamp = sock_flag(sk, SOCK_TSTAMP_NEW); |
9718475e DD |
768 | struct scm_timestamping_internal tss; |
769 | ||
b50a5c70 | 770 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
771 | struct skb_shared_hwtstamps *shhwtstamps = |
772 | skb_hwtstamps(skb); | |
773 | ||
774 | /* Race occurred between timestamp enabling and packet | |
775 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 776 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 777 | __net_timestamp(skb); |
b50a5c70 ML |
778 | false_tstamp = 1; |
779 | } | |
20d49473 PO |
780 | |
781 | if (need_software_tstamp) { | |
782 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
887feae3 DD |
783 | if (new_tstamp) { |
784 | struct __kernel_sock_timeval tv; | |
785 | ||
786 | skb_get_new_timestamp(skb, &tv); | |
787 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_NEW, | |
788 | sizeof(tv), &tv); | |
789 | } else { | |
790 | struct __kernel_old_timeval tv; | |
791 | ||
792 | skb_get_timestamp(skb, &tv); | |
793 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_OLD, | |
794 | sizeof(tv), &tv); | |
795 | } | |
20d49473 | 796 | } else { |
887feae3 DD |
797 | if (new_tstamp) { |
798 | struct __kernel_timespec ts; | |
799 | ||
800 | skb_get_new_timestampns(skb, &ts); | |
801 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_NEW, | |
802 | sizeof(ts), &ts); | |
803 | } else { | |
804 | struct timespec ts; | |
805 | ||
806 | skb_get_timestampns(skb, &ts); | |
807 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_OLD, | |
808 | sizeof(ts), &ts); | |
809 | } | |
20d49473 PO |
810 | } |
811 | } | |
812 | ||
f24b9be5 | 813 | memset(&tss, 0, sizeof(tss)); |
c199105d | 814 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
9718475e | 815 | ktime_to_timespec64_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 816 | empty = 0; |
4d276eb6 | 817 | if (shhwtstamps && |
b9f40e21 | 818 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
b50a5c70 | 819 | !skb_is_swtx_tstamp(skb, false_tstamp) && |
9718475e | 820 | ktime_to_timespec64_cond(shhwtstamps->hwtstamp, tss.ts + 2)) { |
4d276eb6 | 821 | empty = 0; |
aad9c8c4 ML |
822 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && |
823 | !skb_is_err_queue(skb)) | |
824 | put_ts_pktinfo(msg, skb); | |
825 | } | |
1c885808 | 826 | if (!empty) { |
9718475e DD |
827 | if (sock_flag(sk, SOCK_TSTAMP_NEW)) |
828 | put_cmsg_scm_timestamping64(msg, &tss); | |
829 | else | |
830 | put_cmsg_scm_timestamping(msg, &tss); | |
1c885808 | 831 | |
8605330a | 832 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 833 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
834 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
835 | skb->len, skb->data); | |
836 | } | |
92f37fd2 | 837 | } |
7c81fd8b ACM |
838 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
839 | ||
6e3e939f JB |
840 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
841 | struct sk_buff *skb) | |
842 | { | |
843 | int ack; | |
844 | ||
845 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
846 | return; | |
847 | if (!skb->wifi_acked_valid) | |
848 | return; | |
849 | ||
850 | ack = skb->wifi_acked; | |
851 | ||
852 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
853 | } | |
854 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
855 | ||
11165f14 | 856 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
857 | struct sk_buff *skb) | |
3b885787 | 858 | { |
744d5a3e | 859 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 860 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 861 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
862 | } |
863 | ||
767dd033 | 864 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
865 | struct sk_buff *skb) |
866 | { | |
867 | sock_recv_timestamp(msg, sk, skb); | |
868 | sock_recv_drops(msg, sk, skb); | |
869 | } | |
767dd033 | 870 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 871 | |
8c3c447b PA |
872 | INDIRECT_CALLABLE_DECLARE(int inet_recvmsg(struct socket *, struct msghdr *, |
873 | size_t , int )); | |
1b784140 | 874 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 875 | int flags) |
1da177e4 | 876 | { |
8c3c447b PA |
877 | return INDIRECT_CALL_INET4(sock->ops->recvmsg, inet_recvmsg, sock, msg, |
878 | msg_data_left(msg), flags); | |
1da177e4 LT |
879 | } |
880 | ||
85806af0 RD |
881 | /** |
882 | * sock_recvmsg - receive a message from @sock | |
883 | * @sock: socket | |
884 | * @msg: message to receive | |
885 | * @flags: message flags | |
886 | * | |
887 | * Receives @msg from @sock, passing through LSM. Returns the total number | |
888 | * of bytes received, or an error. | |
889 | */ | |
2da62906 | 890 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 891 | { |
2da62906 | 892 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 893 | |
2da62906 | 894 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 895 | } |
c6d409cf | 896 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 897 | |
c1249c0a | 898 | /** |
8a3c245c PT |
899 | * kernel_recvmsg - Receive a message from a socket (kernel space) |
900 | * @sock: The socket to receive the message from | |
901 | * @msg: Received message | |
902 | * @vec: Input s/g array for message data | |
903 | * @num: Size of input s/g array | |
904 | * @size: Number of bytes to read | |
905 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
c1249c0a | 906 | * |
8a3c245c PT |
907 | * On return the msg structure contains the scatter/gather array passed in the |
908 | * vec argument. The array is modified so that it consists of the unfilled | |
909 | * portion of the original array. | |
c1249c0a | 910 | * |
8a3c245c | 911 | * The returned value is the total number of bytes received, or an error. |
c1249c0a | 912 | */ |
8a3c245c | 913 | |
89bddce5 SH |
914 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
915 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
916 | { |
917 | mm_segment_t oldfs = get_fs(); | |
918 | int result; | |
919 | ||
aa563d7b | 920 | iov_iter_kvec(&msg->msg_iter, READ, vec, num, size); |
1da177e4 | 921 | set_fs(KERNEL_DS); |
2da62906 | 922 | result = sock_recvmsg(sock, msg, flags); |
1da177e4 LT |
923 | set_fs(oldfs); |
924 | return result; | |
925 | } | |
c6d409cf | 926 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 927 | |
ce1d4d3e CH |
928 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
929 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 930 | { |
1da177e4 LT |
931 | struct socket *sock; |
932 | int flags; | |
933 | ||
ce1d4d3e CH |
934 | sock = file->private_data; |
935 | ||
35f9c09f ED |
936 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
937 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
938 | flags |= more; | |
ce1d4d3e | 939 | |
e6949583 | 940 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 941 | } |
1da177e4 | 942 | |
9c55e01c | 943 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 944 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
945 | unsigned int flags) |
946 | { | |
947 | struct socket *sock = file->private_data; | |
948 | ||
997b37da | 949 | if (unlikely(!sock->ops->splice_read)) |
95506588 | 950 | return generic_file_splice_read(file, ppos, pipe, len, flags); |
997b37da | 951 | |
9c55e01c JA |
952 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
953 | } | |
954 | ||
8ae5e030 | 955 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 956 | { |
6d652330 AV |
957 | struct file *file = iocb->ki_filp; |
958 | struct socket *sock = file->private_data; | |
0345f931 | 959 | struct msghdr msg = {.msg_iter = *to, |
960 | .msg_iocb = iocb}; | |
8ae5e030 | 961 | ssize_t res; |
ce1d4d3e | 962 | |
8ae5e030 AV |
963 | if (file->f_flags & O_NONBLOCK) |
964 | msg.msg_flags = MSG_DONTWAIT; | |
965 | ||
966 | if (iocb->ki_pos != 0) | |
1da177e4 | 967 | return -ESPIPE; |
027445c3 | 968 | |
66ee59af | 969 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
970 | return 0; |
971 | ||
2da62906 | 972 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
973 | *to = msg.msg_iter; |
974 | return res; | |
1da177e4 LT |
975 | } |
976 | ||
8ae5e030 | 977 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 978 | { |
6d652330 AV |
979 | struct file *file = iocb->ki_filp; |
980 | struct socket *sock = file->private_data; | |
0345f931 | 981 | struct msghdr msg = {.msg_iter = *from, |
982 | .msg_iocb = iocb}; | |
8ae5e030 | 983 | ssize_t res; |
1da177e4 | 984 | |
8ae5e030 | 985 | if (iocb->ki_pos != 0) |
ce1d4d3e | 986 | return -ESPIPE; |
027445c3 | 987 | |
8ae5e030 AV |
988 | if (file->f_flags & O_NONBLOCK) |
989 | msg.msg_flags = MSG_DONTWAIT; | |
990 | ||
6d652330 AV |
991 | if (sock->type == SOCK_SEQPACKET) |
992 | msg.msg_flags |= MSG_EOR; | |
993 | ||
d8725c86 | 994 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
995 | *from = msg.msg_iter; |
996 | return res; | |
1da177e4 LT |
997 | } |
998 | ||
1da177e4 LT |
999 | /* |
1000 | * Atomic setting of ioctl hooks to avoid race | |
1001 | * with module unload. | |
1002 | */ | |
1003 | ||
4a3e2f71 | 1004 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 1005 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 1006 | |
881d966b | 1007 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 1008 | { |
4a3e2f71 | 1009 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 1010 | br_ioctl_hook = hook; |
4a3e2f71 | 1011 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1012 | } |
1013 | EXPORT_SYMBOL(brioctl_set); | |
1014 | ||
4a3e2f71 | 1015 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 1016 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 1017 | |
881d966b | 1018 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 1019 | { |
4a3e2f71 | 1020 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1021 | vlan_ioctl_hook = hook; |
4a3e2f71 | 1022 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
1023 | } |
1024 | EXPORT_SYMBOL(vlan_ioctl_set); | |
1025 | ||
4a3e2f71 | 1026 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 1027 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 1028 | |
89bddce5 | 1029 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 1030 | { |
4a3e2f71 | 1031 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 1032 | dlci_ioctl_hook = hook; |
4a3e2f71 | 1033 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
1034 | } |
1035 | EXPORT_SYMBOL(dlci_ioctl_set); | |
1036 | ||
6b96018b | 1037 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
63ff03ab | 1038 | unsigned int cmd, unsigned long arg) |
6b96018b AB |
1039 | { |
1040 | int err; | |
1041 | void __user *argp = (void __user *)arg; | |
1042 | ||
1043 | err = sock->ops->ioctl(sock, cmd, arg); | |
1044 | ||
1045 | /* | |
1046 | * If this ioctl is unknown try to hand it down | |
1047 | * to the NIC driver. | |
1048 | */ | |
36fd633e AV |
1049 | if (err != -ENOIOCTLCMD) |
1050 | return err; | |
6b96018b | 1051 | |
36fd633e AV |
1052 | if (cmd == SIOCGIFCONF) { |
1053 | struct ifconf ifc; | |
1054 | if (copy_from_user(&ifc, argp, sizeof(struct ifconf))) | |
1055 | return -EFAULT; | |
1056 | rtnl_lock(); | |
1057 | err = dev_ifconf(net, &ifc, sizeof(struct ifreq)); | |
1058 | rtnl_unlock(); | |
1059 | if (!err && copy_to_user(argp, &ifc, sizeof(struct ifconf))) | |
1060 | err = -EFAULT; | |
44c02a2c AV |
1061 | } else { |
1062 | struct ifreq ifr; | |
1063 | bool need_copyout; | |
63ff03ab | 1064 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) |
44c02a2c AV |
1065 | return -EFAULT; |
1066 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1067 | if (!err && need_copyout) | |
63ff03ab | 1068 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) |
44c02a2c | 1069 | return -EFAULT; |
36fd633e | 1070 | } |
6b96018b AB |
1071 | return err; |
1072 | } | |
1073 | ||
1da177e4 LT |
1074 | /* |
1075 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1076 | * what to do with it - that's up to the protocol still. | |
1077 | */ | |
1078 | ||
8a3c245c PT |
1079 | /** |
1080 | * get_net_ns - increment the refcount of the network namespace | |
1081 | * @ns: common namespace (net) | |
1082 | * | |
1083 | * Returns the net's common namespace. | |
1084 | */ | |
1085 | ||
d8d211a2 | 1086 | struct ns_common *get_net_ns(struct ns_common *ns) |
c62cce2c AV |
1087 | { |
1088 | return &get_net(container_of(ns, struct net, ns))->ns; | |
1089 | } | |
d8d211a2 | 1090 | EXPORT_SYMBOL_GPL(get_net_ns); |
c62cce2c | 1091 | |
1da177e4 LT |
1092 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
1093 | { | |
1094 | struct socket *sock; | |
881d966b | 1095 | struct sock *sk; |
1da177e4 LT |
1096 | void __user *argp = (void __user *)arg; |
1097 | int pid, err; | |
881d966b | 1098 | struct net *net; |
1da177e4 | 1099 | |
b69aee04 | 1100 | sock = file->private_data; |
881d966b | 1101 | sk = sock->sk; |
3b1e0a65 | 1102 | net = sock_net(sk); |
44c02a2c AV |
1103 | if (unlikely(cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))) { |
1104 | struct ifreq ifr; | |
1105 | bool need_copyout; | |
1106 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
1107 | return -EFAULT; | |
1108 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1109 | if (!err && need_copyout) | |
1110 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
1111 | return -EFAULT; | |
1da177e4 | 1112 | } else |
3d23e349 | 1113 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1114 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
b1b0c245 | 1115 | err = wext_handle_ioctl(net, cmd, argp); |
1da177e4 | 1116 | } else |
3d23e349 | 1117 | #endif |
89bddce5 | 1118 | switch (cmd) { |
1da177e4 LT |
1119 | case FIOSETOWN: |
1120 | case SIOCSPGRP: | |
1121 | err = -EFAULT; | |
1122 | if (get_user(pid, (int __user *)argp)) | |
1123 | break; | |
393cc3f5 | 1124 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1125 | break; |
1126 | case FIOGETOWN: | |
1127 | case SIOCGPGRP: | |
609d7fa9 | 1128 | err = put_user(f_getown(sock->file), |
89bddce5 | 1129 | (int __user *)argp); |
1da177e4 LT |
1130 | break; |
1131 | case SIOCGIFBR: | |
1132 | case SIOCSIFBR: | |
1133 | case SIOCBRADDBR: | |
1134 | case SIOCBRDELBR: | |
1135 | err = -ENOPKG; | |
1136 | if (!br_ioctl_hook) | |
1137 | request_module("bridge"); | |
1138 | ||
4a3e2f71 | 1139 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1140 | if (br_ioctl_hook) |
881d966b | 1141 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1142 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1143 | break; |
1144 | case SIOCGIFVLAN: | |
1145 | case SIOCSIFVLAN: | |
1146 | err = -ENOPKG; | |
1147 | if (!vlan_ioctl_hook) | |
1148 | request_module("8021q"); | |
1149 | ||
4a3e2f71 | 1150 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1151 | if (vlan_ioctl_hook) |
881d966b | 1152 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1153 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1154 | break; |
1da177e4 LT |
1155 | case SIOCADDDLCI: |
1156 | case SIOCDELDLCI: | |
1157 | err = -ENOPKG; | |
1158 | if (!dlci_ioctl_hook) | |
1159 | request_module("dlci"); | |
1160 | ||
7512cbf6 PE |
1161 | mutex_lock(&dlci_ioctl_mutex); |
1162 | if (dlci_ioctl_hook) | |
1da177e4 | 1163 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1164 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 | 1165 | break; |
c62cce2c AV |
1166 | case SIOCGSKNS: |
1167 | err = -EPERM; | |
1168 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1169 | break; | |
1170 | ||
1171 | err = open_related_ns(&net->ns, get_net_ns); | |
1172 | break; | |
0768e170 AB |
1173 | case SIOCGSTAMP_OLD: |
1174 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
1175 | if (!sock->ops->gettstamp) { |
1176 | err = -ENOIOCTLCMD; | |
1177 | break; | |
1178 | } | |
1179 | err = sock->ops->gettstamp(sock, argp, | |
0768e170 AB |
1180 | cmd == SIOCGSTAMP_OLD, |
1181 | !IS_ENABLED(CONFIG_64BIT)); | |
60747828 | 1182 | break; |
0768e170 AB |
1183 | case SIOCGSTAMP_NEW: |
1184 | case SIOCGSTAMPNS_NEW: | |
1185 | if (!sock->ops->gettstamp) { | |
1186 | err = -ENOIOCTLCMD; | |
1187 | break; | |
1188 | } | |
1189 | err = sock->ops->gettstamp(sock, argp, | |
1190 | cmd == SIOCGSTAMP_NEW, | |
1191 | false); | |
c7cbdbf2 | 1192 | break; |
1da177e4 | 1193 | default: |
63ff03ab | 1194 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1195 | break; |
89bddce5 | 1196 | } |
1da177e4 LT |
1197 | return err; |
1198 | } | |
1199 | ||
8a3c245c PT |
1200 | /** |
1201 | * sock_create_lite - creates a socket | |
1202 | * @family: protocol family (AF_INET, ...) | |
1203 | * @type: communication type (SOCK_STREAM, ...) | |
1204 | * @protocol: protocol (0, ...) | |
1205 | * @res: new socket | |
1206 | * | |
1207 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1208 | * The new socket initialization is not complete, see kernel_accept(). | |
1209 | * Returns 0 or an error. On failure @res is set to %NULL. | |
1210 | * This function internally uses GFP_KERNEL. | |
1211 | */ | |
1212 | ||
1da177e4 LT |
1213 | int sock_create_lite(int family, int type, int protocol, struct socket **res) |
1214 | { | |
1215 | int err; | |
1216 | struct socket *sock = NULL; | |
89bddce5 | 1217 | |
1da177e4 LT |
1218 | err = security_socket_create(family, type, protocol, 1); |
1219 | if (err) | |
1220 | goto out; | |
1221 | ||
1222 | sock = sock_alloc(); | |
1223 | if (!sock) { | |
1224 | err = -ENOMEM; | |
1225 | goto out; | |
1226 | } | |
1227 | ||
1da177e4 | 1228 | sock->type = type; |
7420ed23 VY |
1229 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1230 | if (err) | |
1231 | goto out_release; | |
1232 | ||
1da177e4 LT |
1233 | out: |
1234 | *res = sock; | |
1235 | return err; | |
7420ed23 VY |
1236 | out_release: |
1237 | sock_release(sock); | |
1238 | sock = NULL; | |
1239 | goto out; | |
1da177e4 | 1240 | } |
c6d409cf | 1241 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1242 | |
1243 | /* No kernel lock held - perfect */ | |
ade994f4 | 1244 | static __poll_t sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1245 | { |
3cafb376 | 1246 | struct socket *sock = file->private_data; |
a331de3b | 1247 | __poll_t events = poll_requested_events(wait), flag = 0; |
2d48d67f | 1248 | |
e88958e6 CH |
1249 | if (!sock->ops->poll) |
1250 | return 0; | |
f641f13b | 1251 | |
a331de3b CH |
1252 | if (sk_can_busy_loop(sock->sk)) { |
1253 | /* poll once if requested by the syscall */ | |
1254 | if (events & POLL_BUSY_LOOP) | |
1255 | sk_busy_loop(sock->sk, 1); | |
1256 | ||
1257 | /* if this socket can poll_ll, tell the system call */ | |
1258 | flag = POLL_BUSY_LOOP; | |
1259 | } | |
1260 | ||
1261 | return sock->ops->poll(file, sock, wait) | flag; | |
1da177e4 LT |
1262 | } |
1263 | ||
89bddce5 | 1264 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1265 | { |
b69aee04 | 1266 | struct socket *sock = file->private_data; |
1da177e4 LT |
1267 | |
1268 | return sock->ops->mmap(file, sock, vma); | |
1269 | } | |
1270 | ||
20380731 | 1271 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1272 | { |
6d8c50dc | 1273 | __sock_release(SOCKET_I(inode), inode); |
1da177e4 LT |
1274 | return 0; |
1275 | } | |
1276 | ||
1277 | /* | |
1278 | * Update the socket async list | |
1279 | * | |
1280 | * Fasync_list locking strategy. | |
1281 | * | |
1282 | * 1. fasync_list is modified only under process context socket lock | |
1283 | * i.e. under semaphore. | |
1284 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1285 | * or under socket lock |
1da177e4 LT |
1286 | */ |
1287 | ||
1288 | static int sock_fasync(int fd, struct file *filp, int on) | |
1289 | { | |
989a2979 ED |
1290 | struct socket *sock = filp->private_data; |
1291 | struct sock *sk = sock->sk; | |
eaefd110 | 1292 | struct socket_wq *wq; |
1da177e4 | 1293 | |
989a2979 | 1294 | if (sk == NULL) |
1da177e4 | 1295 | return -EINVAL; |
1da177e4 LT |
1296 | |
1297 | lock_sock(sk); | |
e6476c21 | 1298 | wq = sock->wq; |
eaefd110 | 1299 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1300 | |
eaefd110 | 1301 | if (!wq->fasync_list) |
989a2979 ED |
1302 | sock_reset_flag(sk, SOCK_FASYNC); |
1303 | else | |
bcdce719 | 1304 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1305 | |
989a2979 | 1306 | release_sock(sk); |
1da177e4 LT |
1307 | return 0; |
1308 | } | |
1309 | ||
ceb5d58b | 1310 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1311 | |
ceb5d58b | 1312 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1313 | { |
ceb5d58b | 1314 | if (!wq || !wq->fasync_list) |
1da177e4 | 1315 | return -1; |
ceb5d58b | 1316 | |
89bddce5 | 1317 | switch (how) { |
8d8ad9d7 | 1318 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1319 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1320 | break; |
1321 | goto call_kill; | |
8d8ad9d7 | 1322 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1323 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 LT |
1324 | break; |
1325 | /* fall through */ | |
8d8ad9d7 | 1326 | case SOCK_WAKE_IO: |
89bddce5 | 1327 | call_kill: |
43815482 | 1328 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1329 | break; |
8d8ad9d7 | 1330 | case SOCK_WAKE_URG: |
43815482 | 1331 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1332 | } |
ceb5d58b | 1333 | |
1da177e4 LT |
1334 | return 0; |
1335 | } | |
c6d409cf | 1336 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1337 | |
8a3c245c PT |
1338 | /** |
1339 | * __sock_create - creates a socket | |
1340 | * @net: net namespace | |
1341 | * @family: protocol family (AF_INET, ...) | |
1342 | * @type: communication type (SOCK_STREAM, ...) | |
1343 | * @protocol: protocol (0, ...) | |
1344 | * @res: new socket | |
1345 | * @kern: boolean for kernel space sockets | |
1346 | * | |
1347 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1348 | * Returns 0 or an error. On failure @res is set to %NULL. @kern must | |
1349 | * be set to true if the socket resides in kernel space. | |
1350 | * This function internally uses GFP_KERNEL. | |
1351 | */ | |
1352 | ||
721db93a | 1353 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1354 | struct socket **res, int kern) |
1da177e4 LT |
1355 | { |
1356 | int err; | |
1357 | struct socket *sock; | |
55737fda | 1358 | const struct net_proto_family *pf; |
1da177e4 LT |
1359 | |
1360 | /* | |
89bddce5 | 1361 | * Check protocol is in range |
1da177e4 LT |
1362 | */ |
1363 | if (family < 0 || family >= NPROTO) | |
1364 | return -EAFNOSUPPORT; | |
1365 | if (type < 0 || type >= SOCK_MAX) | |
1366 | return -EINVAL; | |
1367 | ||
1368 | /* Compatibility. | |
1369 | ||
1370 | This uglymoron is moved from INET layer to here to avoid | |
1371 | deadlock in module load. | |
1372 | */ | |
1373 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1374 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1375 | current->comm); | |
1da177e4 LT |
1376 | family = PF_PACKET; |
1377 | } | |
1378 | ||
1379 | err = security_socket_create(family, type, protocol, kern); | |
1380 | if (err) | |
1381 | return err; | |
89bddce5 | 1382 | |
55737fda SH |
1383 | /* |
1384 | * Allocate the socket and allow the family to set things up. if | |
1385 | * the protocol is 0, the family is instructed to select an appropriate | |
1386 | * default. | |
1387 | */ | |
1388 | sock = sock_alloc(); | |
1389 | if (!sock) { | |
e87cc472 | 1390 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1391 | return -ENFILE; /* Not exactly a match, but its the |
1392 | closest posix thing */ | |
1393 | } | |
1394 | ||
1395 | sock->type = type; | |
1396 | ||
95a5afca | 1397 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1398 | /* Attempt to load a protocol module if the find failed. |
1399 | * | |
1400 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1401 | * requested real, full-featured networking support upon configuration. |
1402 | * Otherwise module support will break! | |
1403 | */ | |
190683a9 | 1404 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1405 | request_module("net-pf-%d", family); |
1da177e4 LT |
1406 | #endif |
1407 | ||
55737fda SH |
1408 | rcu_read_lock(); |
1409 | pf = rcu_dereference(net_families[family]); | |
1410 | err = -EAFNOSUPPORT; | |
1411 | if (!pf) | |
1412 | goto out_release; | |
1da177e4 LT |
1413 | |
1414 | /* | |
1415 | * We will call the ->create function, that possibly is in a loadable | |
1416 | * module, so we have to bump that loadable module refcnt first. | |
1417 | */ | |
55737fda | 1418 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1419 | goto out_release; |
1420 | ||
55737fda SH |
1421 | /* Now protected by module ref count */ |
1422 | rcu_read_unlock(); | |
1423 | ||
3f378b68 | 1424 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1425 | if (err < 0) |
1da177e4 | 1426 | goto out_module_put; |
a79af59e | 1427 | |
1da177e4 LT |
1428 | /* |
1429 | * Now to bump the refcnt of the [loadable] module that owns this | |
1430 | * socket at sock_release time we decrement its refcnt. | |
1431 | */ | |
55737fda SH |
1432 | if (!try_module_get(sock->ops->owner)) |
1433 | goto out_module_busy; | |
1434 | ||
1da177e4 LT |
1435 | /* |
1436 | * Now that we're done with the ->create function, the [loadable] | |
1437 | * module can have its refcnt decremented | |
1438 | */ | |
55737fda | 1439 | module_put(pf->owner); |
7420ed23 VY |
1440 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1441 | if (err) | |
3b185525 | 1442 | goto out_sock_release; |
55737fda | 1443 | *res = sock; |
1da177e4 | 1444 | |
55737fda SH |
1445 | return 0; |
1446 | ||
1447 | out_module_busy: | |
1448 | err = -EAFNOSUPPORT; | |
1da177e4 | 1449 | out_module_put: |
55737fda SH |
1450 | sock->ops = NULL; |
1451 | module_put(pf->owner); | |
1452 | out_sock_release: | |
1da177e4 | 1453 | sock_release(sock); |
55737fda SH |
1454 | return err; |
1455 | ||
1456 | out_release: | |
1457 | rcu_read_unlock(); | |
1458 | goto out_sock_release; | |
1da177e4 | 1459 | } |
721db93a | 1460 | EXPORT_SYMBOL(__sock_create); |
1da177e4 | 1461 | |
8a3c245c PT |
1462 | /** |
1463 | * sock_create - creates a socket | |
1464 | * @family: protocol family (AF_INET, ...) | |
1465 | * @type: communication type (SOCK_STREAM, ...) | |
1466 | * @protocol: protocol (0, ...) | |
1467 | * @res: new socket | |
1468 | * | |
1469 | * A wrapper around __sock_create(). | |
1470 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1471 | */ | |
1472 | ||
1da177e4 LT |
1473 | int sock_create(int family, int type, int protocol, struct socket **res) |
1474 | { | |
1b8d7ae4 | 1475 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1476 | } |
c6d409cf | 1477 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1478 | |
8a3c245c PT |
1479 | /** |
1480 | * sock_create_kern - creates a socket (kernel space) | |
1481 | * @net: net namespace | |
1482 | * @family: protocol family (AF_INET, ...) | |
1483 | * @type: communication type (SOCK_STREAM, ...) | |
1484 | * @protocol: protocol (0, ...) | |
1485 | * @res: new socket | |
1486 | * | |
1487 | * A wrapper around __sock_create(). | |
1488 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1489 | */ | |
1490 | ||
eeb1bd5c | 1491 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1492 | { |
eeb1bd5c | 1493 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1494 | } |
c6d409cf | 1495 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1496 | |
9d6a15c3 | 1497 | int __sys_socket(int family, int type, int protocol) |
1da177e4 LT |
1498 | { |
1499 | int retval; | |
1500 | struct socket *sock; | |
a677a039 UD |
1501 | int flags; |
1502 | ||
e38b36f3 UD |
1503 | /* Check the SOCK_* constants for consistency. */ |
1504 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1505 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1506 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1507 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1508 | ||
a677a039 | 1509 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1510 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1511 | return -EINVAL; |
1512 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1513 | |
aaca0bdc UD |
1514 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1515 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1516 | ||
1da177e4 LT |
1517 | retval = sock_create(family, type, protocol, &sock); |
1518 | if (retval < 0) | |
8e1611e2 | 1519 | return retval; |
1da177e4 | 1520 | |
8e1611e2 | 1521 | return sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1522 | } |
1523 | ||
9d6a15c3 DB |
1524 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1525 | { | |
1526 | return __sys_socket(family, type, protocol); | |
1527 | } | |
1528 | ||
1da177e4 LT |
1529 | /* |
1530 | * Create a pair of connected sockets. | |
1531 | */ | |
1532 | ||
6debc8d8 | 1533 | int __sys_socketpair(int family, int type, int protocol, int __user *usockvec) |
1da177e4 LT |
1534 | { |
1535 | struct socket *sock1, *sock2; | |
1536 | int fd1, fd2, err; | |
db349509 | 1537 | struct file *newfile1, *newfile2; |
a677a039 UD |
1538 | int flags; |
1539 | ||
1540 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1541 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1542 | return -EINVAL; |
1543 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1544 | |
aaca0bdc UD |
1545 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1546 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1547 | ||
016a266b AV |
1548 | /* |
1549 | * reserve descriptors and make sure we won't fail | |
1550 | * to return them to userland. | |
1551 | */ | |
1552 | fd1 = get_unused_fd_flags(flags); | |
1553 | if (unlikely(fd1 < 0)) | |
1554 | return fd1; | |
1555 | ||
1556 | fd2 = get_unused_fd_flags(flags); | |
1557 | if (unlikely(fd2 < 0)) { | |
1558 | put_unused_fd(fd1); | |
1559 | return fd2; | |
1560 | } | |
1561 | ||
1562 | err = put_user(fd1, &usockvec[0]); | |
1563 | if (err) | |
1564 | goto out; | |
1565 | ||
1566 | err = put_user(fd2, &usockvec[1]); | |
1567 | if (err) | |
1568 | goto out; | |
1569 | ||
1da177e4 LT |
1570 | /* |
1571 | * Obtain the first socket and check if the underlying protocol | |
1572 | * supports the socketpair call. | |
1573 | */ | |
1574 | ||
1575 | err = sock_create(family, type, protocol, &sock1); | |
016a266b | 1576 | if (unlikely(err < 0)) |
1da177e4 LT |
1577 | goto out; |
1578 | ||
1579 | err = sock_create(family, type, protocol, &sock2); | |
016a266b AV |
1580 | if (unlikely(err < 0)) { |
1581 | sock_release(sock1); | |
1582 | goto out; | |
bf3c23d1 | 1583 | } |
d73aa286 | 1584 | |
d47cd945 DH |
1585 | err = security_socket_socketpair(sock1, sock2); |
1586 | if (unlikely(err)) { | |
1587 | sock_release(sock2); | |
1588 | sock_release(sock1); | |
1589 | goto out; | |
1590 | } | |
1591 | ||
016a266b AV |
1592 | err = sock1->ops->socketpair(sock1, sock2); |
1593 | if (unlikely(err < 0)) { | |
1594 | sock_release(sock2); | |
1595 | sock_release(sock1); | |
1596 | goto out; | |
28407630 AV |
1597 | } |
1598 | ||
aab174f0 | 1599 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1600 | if (IS_ERR(newfile1)) { |
28407630 | 1601 | err = PTR_ERR(newfile1); |
016a266b AV |
1602 | sock_release(sock2); |
1603 | goto out; | |
28407630 AV |
1604 | } |
1605 | ||
aab174f0 | 1606 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1607 | if (IS_ERR(newfile2)) { |
1608 | err = PTR_ERR(newfile2); | |
016a266b AV |
1609 | fput(newfile1); |
1610 | goto out; | |
db349509 AV |
1611 | } |
1612 | ||
157cf649 | 1613 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1614 | |
db349509 AV |
1615 | fd_install(fd1, newfile1); |
1616 | fd_install(fd2, newfile2); | |
d73aa286 | 1617 | return 0; |
1da177e4 | 1618 | |
016a266b | 1619 | out: |
d73aa286 | 1620 | put_unused_fd(fd2); |
d73aa286 | 1621 | put_unused_fd(fd1); |
1da177e4 LT |
1622 | return err; |
1623 | } | |
1624 | ||
6debc8d8 DB |
1625 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1626 | int __user *, usockvec) | |
1627 | { | |
1628 | return __sys_socketpair(family, type, protocol, usockvec); | |
1629 | } | |
1630 | ||
1da177e4 LT |
1631 | /* |
1632 | * Bind a name to a socket. Nothing much to do here since it's | |
1633 | * the protocol's responsibility to handle the local address. | |
1634 | * | |
1635 | * We move the socket address to kernel space before we call | |
1636 | * the protocol layer (having also checked the address is ok). | |
1637 | */ | |
1638 | ||
a87d35d8 | 1639 | int __sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
1da177e4 LT |
1640 | { |
1641 | struct socket *sock; | |
230b1839 | 1642 | struct sockaddr_storage address; |
6cb153ca | 1643 | int err, fput_needed; |
1da177e4 | 1644 | |
89bddce5 | 1645 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1646 | if (sock) { |
43db362d | 1647 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
068b88cc | 1648 | if (!err) { |
89bddce5 | 1649 | err = security_socket_bind(sock, |
230b1839 | 1650 | (struct sockaddr *)&address, |
89bddce5 | 1651 | addrlen); |
6cb153ca BL |
1652 | if (!err) |
1653 | err = sock->ops->bind(sock, | |
89bddce5 | 1654 | (struct sockaddr *) |
230b1839 | 1655 | &address, addrlen); |
1da177e4 | 1656 | } |
6cb153ca | 1657 | fput_light(sock->file, fput_needed); |
89bddce5 | 1658 | } |
1da177e4 LT |
1659 | return err; |
1660 | } | |
1661 | ||
a87d35d8 DB |
1662 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1663 | { | |
1664 | return __sys_bind(fd, umyaddr, addrlen); | |
1665 | } | |
1666 | ||
1da177e4 LT |
1667 | /* |
1668 | * Perform a listen. Basically, we allow the protocol to do anything | |
1669 | * necessary for a listen, and if that works, we mark the socket as | |
1670 | * ready for listening. | |
1671 | */ | |
1672 | ||
25e290ee | 1673 | int __sys_listen(int fd, int backlog) |
1da177e4 LT |
1674 | { |
1675 | struct socket *sock; | |
6cb153ca | 1676 | int err, fput_needed; |
b8e1f9b5 | 1677 | int somaxconn; |
89bddce5 SH |
1678 | |
1679 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1680 | if (sock) { | |
8efa6e93 | 1681 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1682 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1683 | backlog = somaxconn; |
1da177e4 LT |
1684 | |
1685 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1686 | if (!err) |
1687 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1688 | |
6cb153ca | 1689 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1690 | } |
1691 | return err; | |
1692 | } | |
1693 | ||
25e290ee DB |
1694 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1695 | { | |
1696 | return __sys_listen(fd, backlog); | |
1697 | } | |
1698 | ||
1da177e4 LT |
1699 | /* |
1700 | * For accept, we attempt to create a new socket, set up the link | |
1701 | * with the client, wake up the client, then return the new | |
1702 | * connected fd. We collect the address of the connector in kernel | |
1703 | * space and move it to user at the very end. This is unclean because | |
1704 | * we open the socket then return an error. | |
1705 | * | |
1706 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1707 | * status to recvmsg. We need to add that support in a way thats | |
b903036a | 1708 | * clean when we restructure accept also. |
1da177e4 LT |
1709 | */ |
1710 | ||
4541e805 DB |
1711 | int __sys_accept4(int fd, struct sockaddr __user *upeer_sockaddr, |
1712 | int __user *upeer_addrlen, int flags) | |
1da177e4 LT |
1713 | { |
1714 | struct socket *sock, *newsock; | |
39d8c1b6 | 1715 | struct file *newfile; |
6cb153ca | 1716 | int err, len, newfd, fput_needed; |
230b1839 | 1717 | struct sockaddr_storage address; |
1da177e4 | 1718 | |
77d27200 | 1719 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1720 | return -EINVAL; |
1721 | ||
1722 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1723 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1724 | ||
6cb153ca | 1725 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1726 | if (!sock) |
1727 | goto out; | |
1728 | ||
1729 | err = -ENFILE; | |
c6d409cf ED |
1730 | newsock = sock_alloc(); |
1731 | if (!newsock) | |
1da177e4 LT |
1732 | goto out_put; |
1733 | ||
1734 | newsock->type = sock->type; | |
1735 | newsock->ops = sock->ops; | |
1736 | ||
1da177e4 LT |
1737 | /* |
1738 | * We don't need try_module_get here, as the listening socket (sock) | |
1739 | * has the protocol module (sock->ops->owner) held. | |
1740 | */ | |
1741 | __module_get(newsock->ops->owner); | |
1742 | ||
28407630 | 1743 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1744 | if (unlikely(newfd < 0)) { |
1745 | err = newfd; | |
9a1875e6 DM |
1746 | sock_release(newsock); |
1747 | goto out_put; | |
39d8c1b6 | 1748 | } |
aab174f0 | 1749 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1750 | if (IS_ERR(newfile)) { |
28407630 AV |
1751 | err = PTR_ERR(newfile); |
1752 | put_unused_fd(newfd); | |
28407630 AV |
1753 | goto out_put; |
1754 | } | |
39d8c1b6 | 1755 | |
a79af59e FF |
1756 | err = security_socket_accept(sock, newsock); |
1757 | if (err) | |
39d8c1b6 | 1758 | goto out_fd; |
a79af59e | 1759 | |
cdfbabfb | 1760 | err = sock->ops->accept(sock, newsock, sock->file->f_flags, false); |
1da177e4 | 1761 | if (err < 0) |
39d8c1b6 | 1762 | goto out_fd; |
1da177e4 LT |
1763 | |
1764 | if (upeer_sockaddr) { | |
9b2c45d4 DV |
1765 | len = newsock->ops->getname(newsock, |
1766 | (struct sockaddr *)&address, 2); | |
1767 | if (len < 0) { | |
1da177e4 | 1768 | err = -ECONNABORTED; |
39d8c1b6 | 1769 | goto out_fd; |
1da177e4 | 1770 | } |
43db362d | 1771 | err = move_addr_to_user(&address, |
230b1839 | 1772 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1773 | if (err < 0) |
39d8c1b6 | 1774 | goto out_fd; |
1da177e4 LT |
1775 | } |
1776 | ||
1777 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1778 | ||
39d8c1b6 DM |
1779 | fd_install(newfd, newfile); |
1780 | err = newfd; | |
1da177e4 | 1781 | |
1da177e4 | 1782 | out_put: |
6cb153ca | 1783 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1784 | out: |
1785 | return err; | |
39d8c1b6 | 1786 | out_fd: |
9606a216 | 1787 | fput(newfile); |
39d8c1b6 | 1788 | put_unused_fd(newfd); |
1da177e4 LT |
1789 | goto out_put; |
1790 | } | |
1791 | ||
4541e805 DB |
1792 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1793 | int __user *, upeer_addrlen, int, flags) | |
1794 | { | |
1795 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, flags); | |
1796 | } | |
1797 | ||
20f37034 HC |
1798 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1799 | int __user *, upeer_addrlen) | |
aaca0bdc | 1800 | { |
4541e805 | 1801 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1802 | } |
1803 | ||
1da177e4 LT |
1804 | /* |
1805 | * Attempt to connect to a socket with the server address. The address | |
1806 | * is in user space so we verify it is OK and move it to kernel space. | |
1807 | * | |
1808 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1809 | * break bindings | |
1810 | * | |
1811 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1812 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1813 | * include the -EINPROGRESS status for such sockets. | |
1814 | */ | |
1815 | ||
1387c2c2 | 1816 | int __sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
1da177e4 LT |
1817 | { |
1818 | struct socket *sock; | |
230b1839 | 1819 | struct sockaddr_storage address; |
6cb153ca | 1820 | int err, fput_needed; |
1da177e4 | 1821 | |
6cb153ca | 1822 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1823 | if (!sock) |
1824 | goto out; | |
43db362d | 1825 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1826 | if (err < 0) |
1827 | goto out_put; | |
1828 | ||
89bddce5 | 1829 | err = |
230b1839 | 1830 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1831 | if (err) |
1832 | goto out_put; | |
1833 | ||
230b1839 | 1834 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1835 | sock->file->f_flags); |
1836 | out_put: | |
6cb153ca | 1837 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1838 | out: |
1839 | return err; | |
1840 | } | |
1841 | ||
1387c2c2 DB |
1842 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1843 | int, addrlen) | |
1844 | { | |
1845 | return __sys_connect(fd, uservaddr, addrlen); | |
1846 | } | |
1847 | ||
1da177e4 LT |
1848 | /* |
1849 | * Get the local address ('name') of a socket object. Move the obtained | |
1850 | * name to user space. | |
1851 | */ | |
1852 | ||
8882a107 DB |
1853 | int __sys_getsockname(int fd, struct sockaddr __user *usockaddr, |
1854 | int __user *usockaddr_len) | |
1da177e4 LT |
1855 | { |
1856 | struct socket *sock; | |
230b1839 | 1857 | struct sockaddr_storage address; |
9b2c45d4 | 1858 | int err, fput_needed; |
89bddce5 | 1859 | |
6cb153ca | 1860 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1861 | if (!sock) |
1862 | goto out; | |
1863 | ||
1864 | err = security_socket_getsockname(sock); | |
1865 | if (err) | |
1866 | goto out_put; | |
1867 | ||
9b2c45d4 DV |
1868 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 0); |
1869 | if (err < 0) | |
1da177e4 | 1870 | goto out_put; |
9b2c45d4 DV |
1871 | /* "err" is actually length in this case */ |
1872 | err = move_addr_to_user(&address, err, usockaddr, usockaddr_len); | |
1da177e4 LT |
1873 | |
1874 | out_put: | |
6cb153ca | 1875 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1876 | out: |
1877 | return err; | |
1878 | } | |
1879 | ||
8882a107 DB |
1880 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1881 | int __user *, usockaddr_len) | |
1882 | { | |
1883 | return __sys_getsockname(fd, usockaddr, usockaddr_len); | |
1884 | } | |
1885 | ||
1da177e4 LT |
1886 | /* |
1887 | * Get the remote address ('name') of a socket object. Move the obtained | |
1888 | * name to user space. | |
1889 | */ | |
1890 | ||
b21c8f83 DB |
1891 | int __sys_getpeername(int fd, struct sockaddr __user *usockaddr, |
1892 | int __user *usockaddr_len) | |
1da177e4 LT |
1893 | { |
1894 | struct socket *sock; | |
230b1839 | 1895 | struct sockaddr_storage address; |
9b2c45d4 | 1896 | int err, fput_needed; |
1da177e4 | 1897 | |
89bddce5 SH |
1898 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1899 | if (sock != NULL) { | |
1da177e4 LT |
1900 | err = security_socket_getpeername(sock); |
1901 | if (err) { | |
6cb153ca | 1902 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1903 | return err; |
1904 | } | |
1905 | ||
9b2c45d4 DV |
1906 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 1); |
1907 | if (err >= 0) | |
1908 | /* "err" is actually length in this case */ | |
1909 | err = move_addr_to_user(&address, err, usockaddr, | |
89bddce5 | 1910 | usockaddr_len); |
6cb153ca | 1911 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1912 | } |
1913 | return err; | |
1914 | } | |
1915 | ||
b21c8f83 DB |
1916 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1917 | int __user *, usockaddr_len) | |
1918 | { | |
1919 | return __sys_getpeername(fd, usockaddr, usockaddr_len); | |
1920 | } | |
1921 | ||
1da177e4 LT |
1922 | /* |
1923 | * Send a datagram to a given address. We move the address into kernel | |
1924 | * space and check the user space data area is readable before invoking | |
1925 | * the protocol. | |
1926 | */ | |
211b634b DB |
1927 | int __sys_sendto(int fd, void __user *buff, size_t len, unsigned int flags, |
1928 | struct sockaddr __user *addr, int addr_len) | |
1da177e4 LT |
1929 | { |
1930 | struct socket *sock; | |
230b1839 | 1931 | struct sockaddr_storage address; |
1da177e4 LT |
1932 | int err; |
1933 | struct msghdr msg; | |
1934 | struct iovec iov; | |
6cb153ca | 1935 | int fput_needed; |
6cb153ca | 1936 | |
602bd0e9 AV |
1937 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1938 | if (unlikely(err)) | |
1939 | return err; | |
de0fa95c PE |
1940 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1941 | if (!sock) | |
4387ff75 | 1942 | goto out; |
6cb153ca | 1943 | |
89bddce5 | 1944 | msg.msg_name = NULL; |
89bddce5 SH |
1945 | msg.msg_control = NULL; |
1946 | msg.msg_controllen = 0; | |
1947 | msg.msg_namelen = 0; | |
6cb153ca | 1948 | if (addr) { |
43db362d | 1949 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1950 | if (err < 0) |
1951 | goto out_put; | |
230b1839 | 1952 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1953 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1954 | } |
1955 | if (sock->file->f_flags & O_NONBLOCK) | |
1956 | flags |= MSG_DONTWAIT; | |
1957 | msg.msg_flags = flags; | |
d8725c86 | 1958 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 1959 | |
89bddce5 | 1960 | out_put: |
de0fa95c | 1961 | fput_light(sock->file, fput_needed); |
4387ff75 | 1962 | out: |
1da177e4 LT |
1963 | return err; |
1964 | } | |
1965 | ||
211b634b DB |
1966 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
1967 | unsigned int, flags, struct sockaddr __user *, addr, | |
1968 | int, addr_len) | |
1969 | { | |
1970 | return __sys_sendto(fd, buff, len, flags, addr, addr_len); | |
1971 | } | |
1972 | ||
1da177e4 | 1973 | /* |
89bddce5 | 1974 | * Send a datagram down a socket. |
1da177e4 LT |
1975 | */ |
1976 | ||
3e0fa65f | 1977 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1978 | unsigned int, flags) |
1da177e4 | 1979 | { |
211b634b | 1980 | return __sys_sendto(fd, buff, len, flags, NULL, 0); |
1da177e4 LT |
1981 | } |
1982 | ||
1983 | /* | |
89bddce5 | 1984 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1985 | * sender. We verify the buffers are writable and if needed move the |
1986 | * sender address from kernel to user space. | |
1987 | */ | |
7a09e1eb DB |
1988 | int __sys_recvfrom(int fd, void __user *ubuf, size_t size, unsigned int flags, |
1989 | struct sockaddr __user *addr, int __user *addr_len) | |
1da177e4 LT |
1990 | { |
1991 | struct socket *sock; | |
1992 | struct iovec iov; | |
1993 | struct msghdr msg; | |
230b1839 | 1994 | struct sockaddr_storage address; |
89bddce5 | 1995 | int err, err2; |
6cb153ca BL |
1996 | int fput_needed; |
1997 | ||
602bd0e9 AV |
1998 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
1999 | if (unlikely(err)) | |
2000 | return err; | |
de0fa95c | 2001 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 2002 | if (!sock) |
de0fa95c | 2003 | goto out; |
1da177e4 | 2004 | |
89bddce5 SH |
2005 | msg.msg_control = NULL; |
2006 | msg.msg_controllen = 0; | |
f3d33426 HFS |
2007 | /* Save some cycles and don't copy the address if not needed */ |
2008 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
2009 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
2010 | msg.msg_namelen = 0; | |
130ed5d1 | 2011 | msg.msg_iocb = NULL; |
9f138fa6 | 2012 | msg.msg_flags = 0; |
1da177e4 LT |
2013 | if (sock->file->f_flags & O_NONBLOCK) |
2014 | flags |= MSG_DONTWAIT; | |
2da62906 | 2015 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 2016 | |
89bddce5 | 2017 | if (err >= 0 && addr != NULL) { |
43db362d | 2018 | err2 = move_addr_to_user(&address, |
230b1839 | 2019 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
2020 | if (err2 < 0) |
2021 | err = err2; | |
1da177e4 | 2022 | } |
de0fa95c PE |
2023 | |
2024 | fput_light(sock->file, fput_needed); | |
4387ff75 | 2025 | out: |
1da177e4 LT |
2026 | return err; |
2027 | } | |
2028 | ||
7a09e1eb DB |
2029 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
2030 | unsigned int, flags, struct sockaddr __user *, addr, | |
2031 | int __user *, addr_len) | |
2032 | { | |
2033 | return __sys_recvfrom(fd, ubuf, size, flags, addr, addr_len); | |
2034 | } | |
2035 | ||
1da177e4 | 2036 | /* |
89bddce5 | 2037 | * Receive a datagram from a socket. |
1da177e4 LT |
2038 | */ |
2039 | ||
b7c0ddf5 JG |
2040 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
2041 | unsigned int, flags) | |
1da177e4 | 2042 | { |
7a09e1eb | 2043 | return __sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
1da177e4 LT |
2044 | } |
2045 | ||
2046 | /* | |
2047 | * Set a socket option. Because we don't know the option lengths we have | |
2048 | * to pass the user mode parameter for the protocols to sort out. | |
2049 | */ | |
2050 | ||
cc36dca0 DB |
2051 | static int __sys_setsockopt(int fd, int level, int optname, |
2052 | char __user *optval, int optlen) | |
1da177e4 | 2053 | { |
6cb153ca | 2054 | int err, fput_needed; |
1da177e4 LT |
2055 | struct socket *sock; |
2056 | ||
2057 | if (optlen < 0) | |
2058 | return -EINVAL; | |
89bddce5 SH |
2059 | |
2060 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2061 | if (sock != NULL) { | |
2062 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
2063 | if (err) |
2064 | goto out_put; | |
1da177e4 LT |
2065 | |
2066 | if (level == SOL_SOCKET) | |
89bddce5 SH |
2067 | err = |
2068 | sock_setsockopt(sock, level, optname, optval, | |
2069 | optlen); | |
1da177e4 | 2070 | else |
89bddce5 SH |
2071 | err = |
2072 | sock->ops->setsockopt(sock, level, optname, optval, | |
2073 | optlen); | |
6cb153ca BL |
2074 | out_put: |
2075 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2076 | } |
2077 | return err; | |
2078 | } | |
2079 | ||
cc36dca0 DB |
2080 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
2081 | char __user *, optval, int, optlen) | |
2082 | { | |
2083 | return __sys_setsockopt(fd, level, optname, optval, optlen); | |
2084 | } | |
2085 | ||
1da177e4 LT |
2086 | /* |
2087 | * Get a socket option. Because we don't know the option lengths we have | |
2088 | * to pass a user mode parameter for the protocols to sort out. | |
2089 | */ | |
2090 | ||
13a2d70e DB |
2091 | static int __sys_getsockopt(int fd, int level, int optname, |
2092 | char __user *optval, int __user *optlen) | |
1da177e4 | 2093 | { |
6cb153ca | 2094 | int err, fput_needed; |
1da177e4 LT |
2095 | struct socket *sock; |
2096 | ||
89bddce5 SH |
2097 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2098 | if (sock != NULL) { | |
6cb153ca BL |
2099 | err = security_socket_getsockopt(sock, level, optname); |
2100 | if (err) | |
2101 | goto out_put; | |
1da177e4 LT |
2102 | |
2103 | if (level == SOL_SOCKET) | |
89bddce5 SH |
2104 | err = |
2105 | sock_getsockopt(sock, level, optname, optval, | |
2106 | optlen); | |
1da177e4 | 2107 | else |
89bddce5 SH |
2108 | err = |
2109 | sock->ops->getsockopt(sock, level, optname, optval, | |
2110 | optlen); | |
6cb153ca BL |
2111 | out_put: |
2112 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2113 | } |
2114 | return err; | |
2115 | } | |
2116 | ||
13a2d70e DB |
2117 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
2118 | char __user *, optval, int __user *, optlen) | |
2119 | { | |
2120 | return __sys_getsockopt(fd, level, optname, optval, optlen); | |
2121 | } | |
2122 | ||
1da177e4 LT |
2123 | /* |
2124 | * Shutdown a socket. | |
2125 | */ | |
2126 | ||
005a1aea | 2127 | int __sys_shutdown(int fd, int how) |
1da177e4 | 2128 | { |
6cb153ca | 2129 | int err, fput_needed; |
1da177e4 LT |
2130 | struct socket *sock; |
2131 | ||
89bddce5 SH |
2132 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2133 | if (sock != NULL) { | |
1da177e4 | 2134 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
2135 | if (!err) |
2136 | err = sock->ops->shutdown(sock, how); | |
2137 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2138 | } |
2139 | return err; | |
2140 | } | |
2141 | ||
005a1aea DB |
2142 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
2143 | { | |
2144 | return __sys_shutdown(fd, how); | |
2145 | } | |
2146 | ||
89bddce5 | 2147 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
2148 | * fields which are the same type (int / unsigned) on our platforms. |
2149 | */ | |
2150 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
2151 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
2152 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
2153 | ||
c71d8ebe TH |
2154 | struct used_address { |
2155 | struct sockaddr_storage name; | |
2156 | unsigned int name_len; | |
2157 | }; | |
2158 | ||
da184284 AV |
2159 | static int copy_msghdr_from_user(struct msghdr *kmsg, |
2160 | struct user_msghdr __user *umsg, | |
2161 | struct sockaddr __user **save_addr, | |
2162 | struct iovec **iov) | |
1661bf36 | 2163 | { |
ffb07550 | 2164 | struct user_msghdr msg; |
08adb7da AV |
2165 | ssize_t err; |
2166 | ||
ffb07550 | 2167 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 2168 | return -EFAULT; |
dbb490b9 | 2169 | |
864d9664 | 2170 | kmsg->msg_control = (void __force *)msg.msg_control; |
ffb07550 AV |
2171 | kmsg->msg_controllen = msg.msg_controllen; |
2172 | kmsg->msg_flags = msg.msg_flags; | |
2173 | ||
2174 | kmsg->msg_namelen = msg.msg_namelen; | |
2175 | if (!msg.msg_name) | |
6a2a2b3a AS |
2176 | kmsg->msg_namelen = 0; |
2177 | ||
dbb490b9 ML |
2178 | if (kmsg->msg_namelen < 0) |
2179 | return -EINVAL; | |
2180 | ||
1661bf36 | 2181 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 2182 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
2183 | |
2184 | if (save_addr) | |
ffb07550 | 2185 | *save_addr = msg.msg_name; |
08adb7da | 2186 | |
ffb07550 | 2187 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 2188 | if (!save_addr) { |
864d9664 PA |
2189 | err = move_addr_to_kernel(msg.msg_name, |
2190 | kmsg->msg_namelen, | |
08adb7da AV |
2191 | kmsg->msg_name); |
2192 | if (err < 0) | |
2193 | return err; | |
2194 | } | |
2195 | } else { | |
2196 | kmsg->msg_name = NULL; | |
2197 | kmsg->msg_namelen = 0; | |
2198 | } | |
2199 | ||
ffb07550 | 2200 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
2201 | return -EMSGSIZE; |
2202 | ||
0345f931 | 2203 | kmsg->msg_iocb = NULL; |
2204 | ||
ffb07550 AV |
2205 | return import_iovec(save_addr ? READ : WRITE, |
2206 | msg.msg_iov, msg.msg_iovlen, | |
da184284 | 2207 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
1661bf36 DC |
2208 | } |
2209 | ||
666547ff | 2210 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2211 | struct msghdr *msg_sys, unsigned int flags, |
28a94d8f TH |
2212 | struct used_address *used_address, |
2213 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 2214 | { |
89bddce5 SH |
2215 | struct compat_msghdr __user *msg_compat = |
2216 | (struct compat_msghdr __user *)msg; | |
230b1839 | 2217 | struct sockaddr_storage address; |
1da177e4 | 2218 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 2219 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 2220 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 2221 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 2222 | unsigned char *ctl_buf = ctl; |
d8725c86 | 2223 | int ctl_len; |
08adb7da | 2224 | ssize_t err; |
89bddce5 | 2225 | |
08adb7da | 2226 | msg_sys->msg_name = &address; |
1da177e4 | 2227 | |
08449320 | 2228 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2229 | err = get_compat_msghdr(msg_sys, msg_compat, NULL, &iov); |
08449320 | 2230 | else |
08adb7da | 2231 | err = copy_msghdr_from_user(msg_sys, msg, NULL, &iov); |
89bddce5 | 2232 | if (err < 0) |
da184284 | 2233 | return err; |
1da177e4 LT |
2234 | |
2235 | err = -ENOBUFS; | |
2236 | ||
228e548e | 2237 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 2238 | goto out_freeiov; |
28a94d8f | 2239 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2240 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2241 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2242 | err = |
228e548e | 2243 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2244 | sizeof(ctl)); |
1da177e4 LT |
2245 | if (err) |
2246 | goto out_freeiov; | |
228e548e AB |
2247 | ctl_buf = msg_sys->msg_control; |
2248 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2249 | } else if (ctl_len) { |
ac4340fc DM |
2250 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2251 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2252 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2253 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2254 | if (ctl_buf == NULL) |
1da177e4 LT |
2255 | goto out_freeiov; |
2256 | } | |
2257 | err = -EFAULT; | |
2258 | /* | |
228e548e | 2259 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2260 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2261 | * checking falls down on this. | |
2262 | */ | |
fb8621bb | 2263 | if (copy_from_user(ctl_buf, |
228e548e | 2264 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2265 | ctl_len)) |
1da177e4 | 2266 | goto out_freectl; |
228e548e | 2267 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2268 | } |
228e548e | 2269 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2270 | |
2271 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2272 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2273 | /* |
2274 | * If this is sendmmsg() and current destination address is same as | |
2275 | * previously succeeded address, omit asking LSM's decision. | |
2276 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2277 | * destination address never matches. | |
2278 | */ | |
bc909d9d MD |
2279 | if (used_address && msg_sys->msg_name && |
2280 | used_address->name_len == msg_sys->msg_namelen && | |
2281 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2282 | used_address->name_len)) { |
d8725c86 | 2283 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2284 | goto out_freectl; |
2285 | } | |
d8725c86 | 2286 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2287 | /* |
2288 | * If this is sendmmsg() and sending to current destination address was | |
2289 | * successful, remember it. | |
2290 | */ | |
2291 | if (used_address && err >= 0) { | |
2292 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2293 | if (msg_sys->msg_name) |
2294 | memcpy(&used_address->name, msg_sys->msg_name, | |
2295 | used_address->name_len); | |
c71d8ebe | 2296 | } |
1da177e4 LT |
2297 | |
2298 | out_freectl: | |
89bddce5 | 2299 | if (ctl_buf != ctl) |
1da177e4 LT |
2300 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
2301 | out_freeiov: | |
da184284 | 2302 | kfree(iov); |
228e548e AB |
2303 | return err; |
2304 | } | |
2305 | ||
2306 | /* | |
2307 | * BSD sendmsg interface | |
2308 | */ | |
2309 | ||
e1834a32 DB |
2310 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2311 | bool forbid_cmsg_compat) | |
228e548e AB |
2312 | { |
2313 | int fput_needed, err; | |
2314 | struct msghdr msg_sys; | |
1be374a0 AL |
2315 | struct socket *sock; |
2316 | ||
e1834a32 DB |
2317 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2318 | return -EINVAL; | |
2319 | ||
1be374a0 | 2320 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2321 | if (!sock) |
2322 | goto out; | |
2323 | ||
28a94d8f | 2324 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2325 | |
6cb153ca | 2326 | fput_light(sock->file, fput_needed); |
89bddce5 | 2327 | out: |
1da177e4 LT |
2328 | return err; |
2329 | } | |
2330 | ||
666547ff | 2331 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 | 2332 | { |
e1834a32 | 2333 | return __sys_sendmsg(fd, msg, flags, true); |
a7526eb5 AL |
2334 | } |
2335 | ||
228e548e AB |
2336 | /* |
2337 | * Linux sendmmsg interface | |
2338 | */ | |
2339 | ||
2340 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
e1834a32 | 2341 | unsigned int flags, bool forbid_cmsg_compat) |
228e548e AB |
2342 | { |
2343 | int fput_needed, err, datagrams; | |
2344 | struct socket *sock; | |
2345 | struct mmsghdr __user *entry; | |
2346 | struct compat_mmsghdr __user *compat_entry; | |
2347 | struct msghdr msg_sys; | |
c71d8ebe | 2348 | struct used_address used_address; |
f092276d | 2349 | unsigned int oflags = flags; |
228e548e | 2350 | |
e1834a32 DB |
2351 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2352 | return -EINVAL; | |
2353 | ||
98382f41 AB |
2354 | if (vlen > UIO_MAXIOV) |
2355 | vlen = UIO_MAXIOV; | |
228e548e AB |
2356 | |
2357 | datagrams = 0; | |
2358 | ||
2359 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2360 | if (!sock) | |
2361 | return err; | |
2362 | ||
c71d8ebe | 2363 | used_address.name_len = UINT_MAX; |
228e548e AB |
2364 | entry = mmsg; |
2365 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2366 | err = 0; |
f092276d | 2367 | flags |= MSG_BATCH; |
228e548e AB |
2368 | |
2369 | while (datagrams < vlen) { | |
f092276d TH |
2370 | if (datagrams == vlen - 1) |
2371 | flags = oflags; | |
2372 | ||
228e548e | 2373 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2374 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2375 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2376 | if (err < 0) |
2377 | break; | |
2378 | err = __put_user(err, &compat_entry->msg_len); | |
2379 | ++compat_entry; | |
2380 | } else { | |
a7526eb5 | 2381 | err = ___sys_sendmsg(sock, |
666547ff | 2382 | (struct user_msghdr __user *)entry, |
28a94d8f | 2383 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2384 | if (err < 0) |
2385 | break; | |
2386 | err = put_user(err, &entry->msg_len); | |
2387 | ++entry; | |
2388 | } | |
2389 | ||
2390 | if (err) | |
2391 | break; | |
2392 | ++datagrams; | |
3023898b SHY |
2393 | if (msg_data_left(&msg_sys)) |
2394 | break; | |
a78cb84c | 2395 | cond_resched(); |
228e548e AB |
2396 | } |
2397 | ||
228e548e AB |
2398 | fput_light(sock->file, fput_needed); |
2399 | ||
728ffb86 AB |
2400 | /* We only return an error if no datagrams were able to be sent */ |
2401 | if (datagrams != 0) | |
228e548e AB |
2402 | return datagrams; |
2403 | ||
228e548e AB |
2404 | return err; |
2405 | } | |
2406 | ||
2407 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2408 | unsigned int, vlen, unsigned int, flags) | |
2409 | { | |
e1834a32 | 2410 | return __sys_sendmmsg(fd, mmsg, vlen, flags, true); |
228e548e AB |
2411 | } |
2412 | ||
666547ff | 2413 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2414 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2415 | { |
89bddce5 SH |
2416 | struct compat_msghdr __user *msg_compat = |
2417 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2418 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2419 | struct iovec *iov = iovstack; |
1da177e4 | 2420 | unsigned long cmsg_ptr; |
2da62906 | 2421 | int len; |
08adb7da | 2422 | ssize_t err; |
1da177e4 LT |
2423 | |
2424 | /* kernel mode address */ | |
230b1839 | 2425 | struct sockaddr_storage addr; |
1da177e4 LT |
2426 | |
2427 | /* user mode address pointers */ | |
2428 | struct sockaddr __user *uaddr; | |
08adb7da | 2429 | int __user *uaddr_len = COMPAT_NAMELEN(msg); |
89bddce5 | 2430 | |
08adb7da | 2431 | msg_sys->msg_name = &addr; |
1da177e4 | 2432 | |
f3d33426 | 2433 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2434 | err = get_compat_msghdr(msg_sys, msg_compat, &uaddr, &iov); |
f3d33426 | 2435 | else |
08adb7da | 2436 | err = copy_msghdr_from_user(msg_sys, msg, &uaddr, &iov); |
1da177e4 | 2437 | if (err < 0) |
da184284 | 2438 | return err; |
1da177e4 | 2439 | |
a2e27255 ACM |
2440 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2441 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2442 | |
f3d33426 HFS |
2443 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2444 | msg_sys->msg_namelen = 0; | |
2445 | ||
1da177e4 LT |
2446 | if (sock->file->f_flags & O_NONBLOCK) |
2447 | flags |= MSG_DONTWAIT; | |
2da62906 | 2448 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, flags); |
1da177e4 LT |
2449 | if (err < 0) |
2450 | goto out_freeiov; | |
2451 | len = err; | |
2452 | ||
2453 | if (uaddr != NULL) { | |
43db362d | 2454 | err = move_addr_to_user(&addr, |
a2e27255 | 2455 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2456 | uaddr_len); |
1da177e4 LT |
2457 | if (err < 0) |
2458 | goto out_freeiov; | |
2459 | } | |
a2e27255 | 2460 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2461 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2462 | if (err) |
2463 | goto out_freeiov; | |
2464 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2465 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2466 | &msg_compat->msg_controllen); |
2467 | else | |
a2e27255 | 2468 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2469 | &msg->msg_controllen); |
2470 | if (err) | |
2471 | goto out_freeiov; | |
2472 | err = len; | |
2473 | ||
2474 | out_freeiov: | |
da184284 | 2475 | kfree(iov); |
a2e27255 ACM |
2476 | return err; |
2477 | } | |
2478 | ||
2479 | /* | |
2480 | * BSD recvmsg interface | |
2481 | */ | |
2482 | ||
e1834a32 DB |
2483 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2484 | bool forbid_cmsg_compat) | |
a2e27255 ACM |
2485 | { |
2486 | int fput_needed, err; | |
2487 | struct msghdr msg_sys; | |
1be374a0 AL |
2488 | struct socket *sock; |
2489 | ||
e1834a32 DB |
2490 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2491 | return -EINVAL; | |
2492 | ||
1be374a0 | 2493 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2494 | if (!sock) |
2495 | goto out; | |
2496 | ||
a7526eb5 | 2497 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2498 | |
6cb153ca | 2499 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2500 | out: |
2501 | return err; | |
2502 | } | |
2503 | ||
666547ff | 2504 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2505 | unsigned int, flags) |
2506 | { | |
e1834a32 | 2507 | return __sys_recvmsg(fd, msg, flags, true); |
a7526eb5 AL |
2508 | } |
2509 | ||
a2e27255 ACM |
2510 | /* |
2511 | * Linux recvmmsg interface | |
2512 | */ | |
2513 | ||
e11d4284 AB |
2514 | static int do_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2515 | unsigned int vlen, unsigned int flags, | |
2516 | struct timespec64 *timeout) | |
a2e27255 ACM |
2517 | { |
2518 | int fput_needed, err, datagrams; | |
2519 | struct socket *sock; | |
2520 | struct mmsghdr __user *entry; | |
d7256d0e | 2521 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2522 | struct msghdr msg_sys; |
766b9f92 DD |
2523 | struct timespec64 end_time; |
2524 | struct timespec64 timeout64; | |
a2e27255 ACM |
2525 | |
2526 | if (timeout && | |
2527 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2528 | timeout->tv_nsec)) | |
2529 | return -EINVAL; | |
2530 | ||
2531 | datagrams = 0; | |
2532 | ||
2533 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2534 | if (!sock) | |
2535 | return err; | |
2536 | ||
7797dc41 SHY |
2537 | if (likely(!(flags & MSG_ERRQUEUE))) { |
2538 | err = sock_error(sock->sk); | |
2539 | if (err) { | |
2540 | datagrams = err; | |
2541 | goto out_put; | |
2542 | } | |
e623a9e9 | 2543 | } |
a2e27255 ACM |
2544 | |
2545 | entry = mmsg; | |
d7256d0e | 2546 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2547 | |
2548 | while (datagrams < vlen) { | |
2549 | /* | |
2550 | * No need to ask LSM for more than the first datagram. | |
2551 | */ | |
d7256d0e | 2552 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2553 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2554 | &msg_sys, flags & ~MSG_WAITFORONE, |
2555 | datagrams); | |
d7256d0e JMG |
2556 | if (err < 0) |
2557 | break; | |
2558 | err = __put_user(err, &compat_entry->msg_len); | |
2559 | ++compat_entry; | |
2560 | } else { | |
a7526eb5 | 2561 | err = ___sys_recvmsg(sock, |
666547ff | 2562 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2563 | &msg_sys, flags & ~MSG_WAITFORONE, |
2564 | datagrams); | |
d7256d0e JMG |
2565 | if (err < 0) |
2566 | break; | |
2567 | err = put_user(err, &entry->msg_len); | |
2568 | ++entry; | |
2569 | } | |
2570 | ||
a2e27255 ACM |
2571 | if (err) |
2572 | break; | |
a2e27255 ACM |
2573 | ++datagrams; |
2574 | ||
71c5c159 BB |
2575 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2576 | if (flags & MSG_WAITFORONE) | |
2577 | flags |= MSG_DONTWAIT; | |
2578 | ||
a2e27255 | 2579 | if (timeout) { |
766b9f92 | 2580 | ktime_get_ts64(&timeout64); |
c2e6c856 | 2581 | *timeout = timespec64_sub(end_time, timeout64); |
a2e27255 ACM |
2582 | if (timeout->tv_sec < 0) { |
2583 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2584 | break; | |
2585 | } | |
2586 | ||
2587 | /* Timeout, return less than vlen datagrams */ | |
2588 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2589 | break; | |
2590 | } | |
2591 | ||
2592 | /* Out of band data, return right away */ | |
2593 | if (msg_sys.msg_flags & MSG_OOB) | |
2594 | break; | |
a78cb84c | 2595 | cond_resched(); |
a2e27255 ACM |
2596 | } |
2597 | ||
a2e27255 | 2598 | if (err == 0) |
34b88a68 ACM |
2599 | goto out_put; |
2600 | ||
2601 | if (datagrams == 0) { | |
2602 | datagrams = err; | |
2603 | goto out_put; | |
2604 | } | |
a2e27255 | 2605 | |
34b88a68 ACM |
2606 | /* |
2607 | * We may return less entries than requested (vlen) if the | |
2608 | * sock is non block and there aren't enough datagrams... | |
2609 | */ | |
2610 | if (err != -EAGAIN) { | |
a2e27255 | 2611 | /* |
34b88a68 ACM |
2612 | * ... or if recvmsg returns an error after we |
2613 | * received some datagrams, where we record the | |
2614 | * error to return on the next call or if the | |
2615 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2616 | */ |
34b88a68 | 2617 | sock->sk->sk_err = -err; |
a2e27255 | 2618 | } |
34b88a68 ACM |
2619 | out_put: |
2620 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2621 | |
34b88a68 | 2622 | return datagrams; |
a2e27255 ACM |
2623 | } |
2624 | ||
e11d4284 AB |
2625 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2626 | unsigned int vlen, unsigned int flags, | |
2627 | struct __kernel_timespec __user *timeout, | |
2628 | struct old_timespec32 __user *timeout32) | |
a2e27255 ACM |
2629 | { |
2630 | int datagrams; | |
c2e6c856 | 2631 | struct timespec64 timeout_sys; |
a2e27255 | 2632 | |
e11d4284 AB |
2633 | if (timeout && get_timespec64(&timeout_sys, timeout)) |
2634 | return -EFAULT; | |
a2e27255 | 2635 | |
e11d4284 | 2636 | if (timeout32 && get_old_timespec32(&timeout_sys, timeout32)) |
a2e27255 ACM |
2637 | return -EFAULT; |
2638 | ||
e11d4284 AB |
2639 | if (!timeout && !timeout32) |
2640 | return do_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2641 | ||
2642 | datagrams = do_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
a2e27255 | 2643 | |
e11d4284 AB |
2644 | if (datagrams <= 0) |
2645 | return datagrams; | |
2646 | ||
2647 | if (timeout && put_timespec64(&timeout_sys, timeout)) | |
2648 | datagrams = -EFAULT; | |
2649 | ||
2650 | if (timeout32 && put_old_timespec32(&timeout_sys, timeout32)) | |
a2e27255 ACM |
2651 | datagrams = -EFAULT; |
2652 | ||
2653 | return datagrams; | |
2654 | } | |
2655 | ||
1255e269 DB |
2656 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, |
2657 | unsigned int, vlen, unsigned int, flags, | |
c2e6c856 | 2658 | struct __kernel_timespec __user *, timeout) |
1255e269 | 2659 | { |
e11d4284 AB |
2660 | if (flags & MSG_CMSG_COMPAT) |
2661 | return -EINVAL; | |
2662 | ||
2663 | return __sys_recvmmsg(fd, mmsg, vlen, flags, timeout, NULL); | |
2664 | } | |
2665 | ||
2666 | #ifdef CONFIG_COMPAT_32BIT_TIME | |
2667 | SYSCALL_DEFINE5(recvmmsg_time32, int, fd, struct mmsghdr __user *, mmsg, | |
2668 | unsigned int, vlen, unsigned int, flags, | |
2669 | struct old_timespec32 __user *, timeout) | |
2670 | { | |
2671 | if (flags & MSG_CMSG_COMPAT) | |
2672 | return -EINVAL; | |
2673 | ||
2674 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL, timeout); | |
1255e269 | 2675 | } |
e11d4284 | 2676 | #endif |
1255e269 | 2677 | |
a2e27255 | 2678 | #ifdef __ARCH_WANT_SYS_SOCKETCALL |
1da177e4 LT |
2679 | /* Argument list sizes for sys_socketcall */ |
2680 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2681 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2682 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2683 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2684 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2685 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2686 | }; |
2687 | ||
1da177e4 LT |
2688 | #undef AL |
2689 | ||
2690 | /* | |
89bddce5 | 2691 | * System call vectors. |
1da177e4 LT |
2692 | * |
2693 | * Argument checking cleaned up. Saved 20% in size. | |
2694 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2695 | * it is set by the callees. |
1da177e4 LT |
2696 | */ |
2697 | ||
3e0fa65f | 2698 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2699 | { |
2950fa9d | 2700 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2701 | unsigned long a0, a1; |
1da177e4 | 2702 | int err; |
47379052 | 2703 | unsigned int len; |
1da177e4 | 2704 | |
228e548e | 2705 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 | 2706 | return -EINVAL; |
c8e8cd57 | 2707 | call = array_index_nospec(call, SYS_SENDMMSG + 1); |
1da177e4 | 2708 | |
47379052 AV |
2709 | len = nargs[call]; |
2710 | if (len > sizeof(a)) | |
2711 | return -EINVAL; | |
2712 | ||
1da177e4 | 2713 | /* copy_from_user should be SMP safe. */ |
47379052 | 2714 | if (copy_from_user(a, args, len)) |
1da177e4 | 2715 | return -EFAULT; |
3ec3b2fb | 2716 | |
2950fa9d CG |
2717 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2718 | if (err) | |
2719 | return err; | |
3ec3b2fb | 2720 | |
89bddce5 SH |
2721 | a0 = a[0]; |
2722 | a1 = a[1]; | |
2723 | ||
2724 | switch (call) { | |
2725 | case SYS_SOCKET: | |
9d6a15c3 | 2726 | err = __sys_socket(a0, a1, a[2]); |
89bddce5 SH |
2727 | break; |
2728 | case SYS_BIND: | |
a87d35d8 | 2729 | err = __sys_bind(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2730 | break; |
2731 | case SYS_CONNECT: | |
1387c2c2 | 2732 | err = __sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2733 | break; |
2734 | case SYS_LISTEN: | |
25e290ee | 2735 | err = __sys_listen(a0, a1); |
89bddce5 SH |
2736 | break; |
2737 | case SYS_ACCEPT: | |
4541e805 DB |
2738 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2739 | (int __user *)a[2], 0); | |
89bddce5 SH |
2740 | break; |
2741 | case SYS_GETSOCKNAME: | |
2742 | err = | |
8882a107 DB |
2743 | __sys_getsockname(a0, (struct sockaddr __user *)a1, |
2744 | (int __user *)a[2]); | |
89bddce5 SH |
2745 | break; |
2746 | case SYS_GETPEERNAME: | |
2747 | err = | |
b21c8f83 DB |
2748 | __sys_getpeername(a0, (struct sockaddr __user *)a1, |
2749 | (int __user *)a[2]); | |
89bddce5 SH |
2750 | break; |
2751 | case SYS_SOCKETPAIR: | |
6debc8d8 | 2752 | err = __sys_socketpair(a0, a1, a[2], (int __user *)a[3]); |
89bddce5 SH |
2753 | break; |
2754 | case SYS_SEND: | |
f3bf896b DB |
2755 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2756 | NULL, 0); | |
89bddce5 SH |
2757 | break; |
2758 | case SYS_SENDTO: | |
211b634b DB |
2759 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2760 | (struct sockaddr __user *)a[4], a[5]); | |
89bddce5 SH |
2761 | break; |
2762 | case SYS_RECV: | |
d27e9afc DB |
2763 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2764 | NULL, NULL); | |
89bddce5 SH |
2765 | break; |
2766 | case SYS_RECVFROM: | |
7a09e1eb DB |
2767 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2768 | (struct sockaddr __user *)a[4], | |
2769 | (int __user *)a[5]); | |
89bddce5 SH |
2770 | break; |
2771 | case SYS_SHUTDOWN: | |
005a1aea | 2772 | err = __sys_shutdown(a0, a1); |
89bddce5 SH |
2773 | break; |
2774 | case SYS_SETSOCKOPT: | |
cc36dca0 DB |
2775 | err = __sys_setsockopt(a0, a1, a[2], (char __user *)a[3], |
2776 | a[4]); | |
89bddce5 SH |
2777 | break; |
2778 | case SYS_GETSOCKOPT: | |
2779 | err = | |
13a2d70e DB |
2780 | __sys_getsockopt(a0, a1, a[2], (char __user *)a[3], |
2781 | (int __user *)a[4]); | |
89bddce5 SH |
2782 | break; |
2783 | case SYS_SENDMSG: | |
e1834a32 DB |
2784 | err = __sys_sendmsg(a0, (struct user_msghdr __user *)a1, |
2785 | a[2], true); | |
89bddce5 | 2786 | break; |
228e548e | 2787 | case SYS_SENDMMSG: |
e1834a32 DB |
2788 | err = __sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
2789 | a[3], true); | |
228e548e | 2790 | break; |
89bddce5 | 2791 | case SYS_RECVMSG: |
e1834a32 DB |
2792 | err = __sys_recvmsg(a0, (struct user_msghdr __user *)a1, |
2793 | a[2], true); | |
89bddce5 | 2794 | break; |
a2e27255 | 2795 | case SYS_RECVMMSG: |
e11d4284 AB |
2796 | if (IS_ENABLED(CONFIG_64BIT) || !IS_ENABLED(CONFIG_64BIT_TIME)) |
2797 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, | |
2798 | a[2], a[3], | |
2799 | (struct __kernel_timespec __user *)a[4], | |
2800 | NULL); | |
2801 | else | |
2802 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, | |
2803 | a[2], a[3], NULL, | |
2804 | (struct old_timespec32 __user *)a[4]); | |
a2e27255 | 2805 | break; |
de11defe | 2806 | case SYS_ACCEPT4: |
4541e805 DB |
2807 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2808 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2809 | break; |
89bddce5 SH |
2810 | default: |
2811 | err = -EINVAL; | |
2812 | break; | |
1da177e4 LT |
2813 | } |
2814 | return err; | |
2815 | } | |
2816 | ||
89bddce5 | 2817 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2818 | |
55737fda SH |
2819 | /** |
2820 | * sock_register - add a socket protocol handler | |
2821 | * @ops: description of protocol | |
2822 | * | |
1da177e4 LT |
2823 | * This function is called by a protocol handler that wants to |
2824 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2825 | * socket interface. The value ops->family corresponds to the |
55737fda | 2826 | * socket system call protocol family. |
1da177e4 | 2827 | */ |
f0fd27d4 | 2828 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2829 | { |
2830 | int err; | |
2831 | ||
2832 | if (ops->family >= NPROTO) { | |
3410f22e | 2833 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2834 | return -ENOBUFS; |
2835 | } | |
55737fda SH |
2836 | |
2837 | spin_lock(&net_family_lock); | |
190683a9 ED |
2838 | if (rcu_dereference_protected(net_families[ops->family], |
2839 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2840 | err = -EEXIST; |
2841 | else { | |
cf778b00 | 2842 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2843 | err = 0; |
2844 | } | |
55737fda SH |
2845 | spin_unlock(&net_family_lock); |
2846 | ||
3410f22e | 2847 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2848 | return err; |
2849 | } | |
c6d409cf | 2850 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2851 | |
55737fda SH |
2852 | /** |
2853 | * sock_unregister - remove a protocol handler | |
2854 | * @family: protocol family to remove | |
2855 | * | |
1da177e4 LT |
2856 | * This function is called by a protocol handler that wants to |
2857 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2858 | * new socket creation. |
2859 | * | |
2860 | * If protocol handler is a module, then it can use module reference | |
2861 | * counts to protect against new references. If protocol handler is not | |
2862 | * a module then it needs to provide its own protection in | |
2863 | * the ops->create routine. | |
1da177e4 | 2864 | */ |
f0fd27d4 | 2865 | void sock_unregister(int family) |
1da177e4 | 2866 | { |
f0fd27d4 | 2867 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2868 | |
55737fda | 2869 | spin_lock(&net_family_lock); |
a9b3cd7f | 2870 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2871 | spin_unlock(&net_family_lock); |
2872 | ||
2873 | synchronize_rcu(); | |
2874 | ||
3410f22e | 2875 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2876 | } |
c6d409cf | 2877 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2878 | |
bf2ae2e4 XL |
2879 | bool sock_is_registered(int family) |
2880 | { | |
66b51b0a | 2881 | return family < NPROTO && rcu_access_pointer(net_families[family]); |
bf2ae2e4 XL |
2882 | } |
2883 | ||
77d76ea3 | 2884 | static int __init sock_init(void) |
1da177e4 | 2885 | { |
b3e19d92 | 2886 | int err; |
2ca794e5 EB |
2887 | /* |
2888 | * Initialize the network sysctl infrastructure. | |
2889 | */ | |
2890 | err = net_sysctl_init(); | |
2891 | if (err) | |
2892 | goto out; | |
b3e19d92 | 2893 | |
1da177e4 | 2894 | /* |
89bddce5 | 2895 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2896 | */ |
2897 | skb_init(); | |
1da177e4 LT |
2898 | |
2899 | /* | |
89bddce5 | 2900 | * Initialize the protocols module. |
1da177e4 LT |
2901 | */ |
2902 | ||
2903 | init_inodecache(); | |
b3e19d92 NP |
2904 | |
2905 | err = register_filesystem(&sock_fs_type); | |
2906 | if (err) | |
2907 | goto out_fs; | |
1da177e4 | 2908 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2909 | if (IS_ERR(sock_mnt)) { |
2910 | err = PTR_ERR(sock_mnt); | |
2911 | goto out_mount; | |
2912 | } | |
77d76ea3 AK |
2913 | |
2914 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2915 | */ |
2916 | ||
2917 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
2918 | err = netfilter_init(); |
2919 | if (err) | |
2920 | goto out; | |
1da177e4 | 2921 | #endif |
cbeb321a | 2922 | |
408eccce | 2923 | ptp_classifier_init(); |
c1f19b51 | 2924 | |
b3e19d92 NP |
2925 | out: |
2926 | return err; | |
2927 | ||
2928 | out_mount: | |
2929 | unregister_filesystem(&sock_fs_type); | |
2930 | out_fs: | |
2931 | goto out; | |
1da177e4 LT |
2932 | } |
2933 | ||
77d76ea3 AK |
2934 | core_initcall(sock_init); /* early initcall */ |
2935 | ||
1da177e4 LT |
2936 | #ifdef CONFIG_PROC_FS |
2937 | void socket_seq_show(struct seq_file *seq) | |
2938 | { | |
648845ab TZ |
2939 | seq_printf(seq, "sockets: used %d\n", |
2940 | sock_inuse_get(seq->private)); | |
1da177e4 | 2941 | } |
89bddce5 | 2942 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 2943 | |
89bbfc95 | 2944 | #ifdef CONFIG_COMPAT |
36fd633e | 2945 | static int compat_dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 2946 | { |
6b96018b | 2947 | struct compat_ifconf ifc32; |
7a229387 | 2948 | struct ifconf ifc; |
7a229387 AB |
2949 | int err; |
2950 | ||
6b96018b | 2951 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2952 | return -EFAULT; |
2953 | ||
36fd633e AV |
2954 | ifc.ifc_len = ifc32.ifc_len; |
2955 | ifc.ifc_req = compat_ptr(ifc32.ifcbuf); | |
7a229387 | 2956 | |
36fd633e AV |
2957 | rtnl_lock(); |
2958 | err = dev_ifconf(net, &ifc, sizeof(struct compat_ifreq)); | |
2959 | rtnl_unlock(); | |
7a229387 AB |
2960 | if (err) |
2961 | return err; | |
2962 | ||
36fd633e | 2963 | ifc32.ifc_len = ifc.ifc_len; |
6b96018b | 2964 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2965 | return -EFAULT; |
2966 | ||
2967 | return 0; | |
2968 | } | |
2969 | ||
6b96018b | 2970 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 2971 | { |
3a7da39d BH |
2972 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
2973 | bool convert_in = false, convert_out = false; | |
44c02a2c AV |
2974 | size_t buf_size = 0; |
2975 | struct ethtool_rxnfc __user *rxnfc = NULL; | |
2976 | struct ifreq ifr; | |
3a7da39d BH |
2977 | u32 rule_cnt = 0, actual_rule_cnt; |
2978 | u32 ethcmd; | |
7a229387 | 2979 | u32 data; |
3a7da39d | 2980 | int ret; |
7a229387 | 2981 | |
3a7da39d BH |
2982 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
2983 | return -EFAULT; | |
7a229387 | 2984 | |
3a7da39d BH |
2985 | compat_rxnfc = compat_ptr(data); |
2986 | ||
2987 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
2988 | return -EFAULT; |
2989 | ||
3a7da39d BH |
2990 | /* Most ethtool structures are defined without padding. |
2991 | * Unfortunately struct ethtool_rxnfc is an exception. | |
2992 | */ | |
2993 | switch (ethcmd) { | |
2994 | default: | |
2995 | break; | |
2996 | case ETHTOOL_GRXCLSRLALL: | |
2997 | /* Buffer size is variable */ | |
2998 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
2999 | return -EFAULT; | |
3000 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
3001 | return -ENOMEM; | |
3002 | buf_size += rule_cnt * sizeof(u32); | |
3003 | /* fall through */ | |
3004 | case ETHTOOL_GRXRINGS: | |
3005 | case ETHTOOL_GRXCLSRLCNT: | |
3006 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 3007 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
3008 | convert_out = true; |
3009 | /* fall through */ | |
3010 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
3011 | buf_size += sizeof(struct ethtool_rxnfc); |
3012 | convert_in = true; | |
44c02a2c | 3013 | rxnfc = compat_alloc_user_space(buf_size); |
3a7da39d BH |
3014 | break; |
3015 | } | |
3016 | ||
44c02a2c | 3017 | if (copy_from_user(&ifr.ifr_name, &ifr32->ifr_name, IFNAMSIZ)) |
7a229387 AB |
3018 | return -EFAULT; |
3019 | ||
44c02a2c | 3020 | ifr.ifr_data = convert_in ? rxnfc : (void __user *)compat_rxnfc; |
7a229387 | 3021 | |
3a7da39d | 3022 | if (convert_in) { |
127fe533 | 3023 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
3024 | * fs.ring_cookie and at the end of fs, but nowhere else. |
3025 | */ | |
127fe533 AD |
3026 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
3027 | sizeof(compat_rxnfc->fs.m_ext) != | |
3028 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
3029 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
3030 | BUILD_BUG_ON( |
3031 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
3032 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
3033 | offsetof(struct ethtool_rxnfc, fs.location) - | |
3034 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
3035 | ||
3036 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
3037 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
3038 | (void __user *)rxnfc) || | |
3a7da39d BH |
3039 | copy_in_user(&rxnfc->fs.ring_cookie, |
3040 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 | 3041 | (void __user *)(&rxnfc->fs.location + 1) - |
b6168562 WW |
3042 | (void __user *)&rxnfc->fs.ring_cookie)) |
3043 | return -EFAULT; | |
3044 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3045 | if (put_user(rule_cnt, &rxnfc->rule_cnt)) | |
3046 | return -EFAULT; | |
3047 | } else if (copy_in_user(&rxnfc->rule_cnt, | |
3048 | &compat_rxnfc->rule_cnt, | |
3049 | sizeof(rxnfc->rule_cnt))) | |
3a7da39d BH |
3050 | return -EFAULT; |
3051 | } | |
3052 | ||
44c02a2c | 3053 | ret = dev_ioctl(net, SIOCETHTOOL, &ifr, NULL); |
3a7da39d BH |
3054 | if (ret) |
3055 | return ret; | |
3056 | ||
3057 | if (convert_out) { | |
3058 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
3059 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
3060 | (const void __user *)rxnfc) || | |
3a7da39d BH |
3061 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
3062 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
3063 | (const void __user *)(&rxnfc->fs.location + 1) - |
3064 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
3065 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
3066 | sizeof(rxnfc->rule_cnt))) | |
3067 | return -EFAULT; | |
3068 | ||
3069 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3070 | /* As an optimisation, we only copy the actual | |
3071 | * number of rules that the underlying | |
3072 | * function returned. Since Mallory might | |
3073 | * change the rule count in user memory, we | |
3074 | * check that it is less than the rule count | |
3075 | * originally given (as the user buffer size), | |
3076 | * which has been range-checked. | |
3077 | */ | |
3078 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
3079 | return -EFAULT; | |
3080 | if (actual_rule_cnt < rule_cnt) | |
3081 | rule_cnt = actual_rule_cnt; | |
3082 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
3083 | &rxnfc->rule_locs[0], | |
3084 | rule_cnt * sizeof(u32))) | |
3085 | return -EFAULT; | |
3086 | } | |
3087 | } | |
3088 | ||
3089 | return 0; | |
7a229387 AB |
3090 | } |
3091 | ||
7a50a240 AB |
3092 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
3093 | { | |
7a50a240 | 3094 | compat_uptr_t uptr32; |
44c02a2c AV |
3095 | struct ifreq ifr; |
3096 | void __user *saved; | |
3097 | int err; | |
7a50a240 | 3098 | |
44c02a2c | 3099 | if (copy_from_user(&ifr, uifr32, sizeof(struct compat_ifreq))) |
7a50a240 AB |
3100 | return -EFAULT; |
3101 | ||
3102 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
3103 | return -EFAULT; | |
3104 | ||
44c02a2c AV |
3105 | saved = ifr.ifr_settings.ifs_ifsu.raw_hdlc; |
3106 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = compat_ptr(uptr32); | |
7a229387 | 3107 | |
44c02a2c AV |
3108 | err = dev_ioctl(net, SIOCWANDEV, &ifr, NULL); |
3109 | if (!err) { | |
3110 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = saved; | |
3111 | if (copy_to_user(uifr32, &ifr, sizeof(struct compat_ifreq))) | |
3112 | err = -EFAULT; | |
ccbd6a5a | 3113 | } |
44c02a2c | 3114 | return err; |
7a229387 AB |
3115 | } |
3116 | ||
590d4693 BH |
3117 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
3118 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 3119 | struct compat_ifreq __user *u_ifreq32) |
7a229387 | 3120 | { |
44c02a2c | 3121 | struct ifreq ifreq; |
7a229387 AB |
3122 | u32 data32; |
3123 | ||
44c02a2c | 3124 | if (copy_from_user(ifreq.ifr_name, u_ifreq32->ifr_name, IFNAMSIZ)) |
7a229387 | 3125 | return -EFAULT; |
44c02a2c | 3126 | if (get_user(data32, &u_ifreq32->ifr_data)) |
7a229387 | 3127 | return -EFAULT; |
44c02a2c | 3128 | ifreq.ifr_data = compat_ptr(data32); |
7a229387 | 3129 | |
44c02a2c | 3130 | return dev_ioctl(net, cmd, &ifreq, NULL); |
7a229387 AB |
3131 | } |
3132 | ||
37ac39bd JB |
3133 | static int compat_ifreq_ioctl(struct net *net, struct socket *sock, |
3134 | unsigned int cmd, | |
3135 | struct compat_ifreq __user *uifr32) | |
3136 | { | |
3137 | struct ifreq __user *uifr; | |
3138 | int err; | |
3139 | ||
3140 | /* Handle the fact that while struct ifreq has the same *layout* on | |
3141 | * 32/64 for everything but ifreq::ifru_ifmap and ifreq::ifru_data, | |
3142 | * which are handled elsewhere, it still has different *size* due to | |
3143 | * ifreq::ifru_ifmap (which is 16 bytes on 32 bit, 24 bytes on 64-bit, | |
3144 | * resulting in struct ifreq being 32 and 40 bytes respectively). | |
3145 | * As a result, if the struct happens to be at the end of a page and | |
3146 | * the next page isn't readable/writable, we get a fault. To prevent | |
3147 | * that, copy back and forth to the full size. | |
3148 | */ | |
3149 | ||
3150 | uifr = compat_alloc_user_space(sizeof(*uifr)); | |
3151 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
3152 | return -EFAULT; | |
3153 | ||
3154 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
3155 | ||
3156 | if (!err) { | |
3157 | switch (cmd) { | |
3158 | case SIOCGIFFLAGS: | |
3159 | case SIOCGIFMETRIC: | |
3160 | case SIOCGIFMTU: | |
3161 | case SIOCGIFMEM: | |
3162 | case SIOCGIFHWADDR: | |
3163 | case SIOCGIFINDEX: | |
3164 | case SIOCGIFADDR: | |
3165 | case SIOCGIFBRDADDR: | |
3166 | case SIOCGIFDSTADDR: | |
3167 | case SIOCGIFNETMASK: | |
3168 | case SIOCGIFPFLAGS: | |
3169 | case SIOCGIFTXQLEN: | |
3170 | case SIOCGMIIPHY: | |
3171 | case SIOCGMIIREG: | |
c6c9fee3 | 3172 | case SIOCGIFNAME: |
37ac39bd JB |
3173 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
3174 | err = -EFAULT; | |
3175 | break; | |
3176 | } | |
3177 | } | |
3178 | return err; | |
3179 | } | |
3180 | ||
a2116ed2 AB |
3181 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
3182 | struct compat_ifreq __user *uifr32) | |
3183 | { | |
3184 | struct ifreq ifr; | |
3185 | struct compat_ifmap __user *uifmap32; | |
a2116ed2 AB |
3186 | int err; |
3187 | ||
3188 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
3189 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3190 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3191 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3192 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3193 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3194 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3195 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3196 | if (err) |
3197 | return -EFAULT; | |
3198 | ||
44c02a2c | 3199 | err = dev_ioctl(net, cmd, &ifr, NULL); |
a2116ed2 AB |
3200 | |
3201 | if (cmd == SIOCGIFMAP && !err) { | |
3202 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3203 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3204 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3205 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3206 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3207 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3208 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3209 | if (err) |
3210 | err = -EFAULT; | |
3211 | } | |
3212 | return err; | |
3213 | } | |
3214 | ||
7a229387 | 3215 | struct rtentry32 { |
c6d409cf | 3216 | u32 rt_pad1; |
7a229387 AB |
3217 | struct sockaddr rt_dst; /* target address */ |
3218 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
3219 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
3220 | unsigned short rt_flags; |
3221 | short rt_pad2; | |
3222 | u32 rt_pad3; | |
3223 | unsigned char rt_tos; | |
3224 | unsigned char rt_class; | |
3225 | short rt_pad4; | |
3226 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 3227 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
3228 | u32 rt_mtu; /* per route MTU/Window */ |
3229 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
3230 | unsigned short rt_irtt; /* Initial RTT */ |
3231 | }; | |
3232 | ||
3233 | struct in6_rtmsg32 { | |
3234 | struct in6_addr rtmsg_dst; | |
3235 | struct in6_addr rtmsg_src; | |
3236 | struct in6_addr rtmsg_gateway; | |
3237 | u32 rtmsg_type; | |
3238 | u16 rtmsg_dst_len; | |
3239 | u16 rtmsg_src_len; | |
3240 | u32 rtmsg_metric; | |
3241 | u32 rtmsg_info; | |
3242 | u32 rtmsg_flags; | |
3243 | s32 rtmsg_ifindex; | |
3244 | }; | |
3245 | ||
6b96018b AB |
3246 | static int routing_ioctl(struct net *net, struct socket *sock, |
3247 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3248 | { |
3249 | int ret; | |
3250 | void *r = NULL; | |
3251 | struct in6_rtmsg r6; | |
3252 | struct rtentry r4; | |
3253 | char devname[16]; | |
3254 | u32 rtdev; | |
3255 | mm_segment_t old_fs = get_fs(); | |
3256 | ||
6b96018b AB |
3257 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3258 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3259 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3260 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
3261 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3262 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3263 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3264 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3265 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3266 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3267 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3268 | |
3269 | r = (void *) &r6; | |
3270 | } else { /* ipv4 */ | |
6b96018b | 3271 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3272 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3273 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
3274 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
3275 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3276 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3277 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
3278 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3279 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3280 | if (rtdev) { |
c6d409cf | 3281 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3282 | r4.rt_dev = (char __user __force *)devname; |
3283 | devname[15] = 0; | |
7a229387 AB |
3284 | } else |
3285 | r4.rt_dev = NULL; | |
3286 | ||
3287 | r = (void *) &r4; | |
3288 | } | |
3289 | ||
3290 | if (ret) { | |
3291 | ret = -EFAULT; | |
3292 | goto out; | |
3293 | } | |
3294 | ||
c6d409cf | 3295 | set_fs(KERNEL_DS); |
63ff03ab | 3296 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3297 | set_fs(old_fs); |
7a229387 AB |
3298 | |
3299 | out: | |
7a229387 AB |
3300 | return ret; |
3301 | } | |
3302 | ||
3303 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3304 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3305 | * use compatible ioctls |
7a229387 | 3306 | */ |
6b96018b | 3307 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3308 | { |
6b96018b | 3309 | compat_ulong_t tmp; |
7a229387 | 3310 | |
6b96018b | 3311 | if (get_user(tmp, argp)) |
7a229387 AB |
3312 | return -EFAULT; |
3313 | if (tmp == BRCTL_GET_VERSION) | |
3314 | return BRCTL_VERSION + 1; | |
3315 | return -EINVAL; | |
3316 | } | |
3317 | ||
6b96018b AB |
3318 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3319 | unsigned int cmd, unsigned long arg) | |
3320 | { | |
3321 | void __user *argp = compat_ptr(arg); | |
3322 | struct sock *sk = sock->sk; | |
3323 | struct net *net = sock_net(sk); | |
7a229387 | 3324 | |
6b96018b | 3325 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3326 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3327 | |
3328 | switch (cmd) { | |
3329 | case SIOCSIFBR: | |
3330 | case SIOCGIFBR: | |
3331 | return old_bridge_ioctl(argp); | |
6b96018b | 3332 | case SIOCGIFCONF: |
36fd633e | 3333 | return compat_dev_ifconf(net, argp); |
6b96018b AB |
3334 | case SIOCETHTOOL: |
3335 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3336 | case SIOCWANDEV: |
3337 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3338 | case SIOCGIFMAP: |
3339 | case SIOCSIFMAP: | |
3340 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3341 | case SIOCADDRT: |
3342 | case SIOCDELRT: | |
3343 | return routing_ioctl(net, sock, cmd, argp); | |
0768e170 AB |
3344 | case SIOCGSTAMP_OLD: |
3345 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
3346 | if (!sock->ops->gettstamp) |
3347 | return -ENOIOCTLCMD; | |
0768e170 | 3348 | return sock->ops->gettstamp(sock, argp, cmd == SIOCGSTAMP_OLD, |
c7cbdbf2 AB |
3349 | !COMPAT_USE_64BIT_TIME); |
3350 | ||
590d4693 BH |
3351 | case SIOCBONDSLAVEINFOQUERY: |
3352 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3353 | case SIOCSHWTSTAMP: |
fd468c74 | 3354 | case SIOCGHWTSTAMP: |
590d4693 | 3355 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3356 | |
3357 | case FIOSETOWN: | |
3358 | case SIOCSPGRP: | |
3359 | case FIOGETOWN: | |
3360 | case SIOCGPGRP: | |
3361 | case SIOCBRADDBR: | |
3362 | case SIOCBRDELBR: | |
3363 | case SIOCGIFVLAN: | |
3364 | case SIOCSIFVLAN: | |
3365 | case SIOCADDDLCI: | |
3366 | case SIOCDELDLCI: | |
c62cce2c | 3367 | case SIOCGSKNS: |
0768e170 AB |
3368 | case SIOCGSTAMP_NEW: |
3369 | case SIOCGSTAMPNS_NEW: | |
6b96018b AB |
3370 | return sock_ioctl(file, cmd, arg); |
3371 | ||
3372 | case SIOCGIFFLAGS: | |
3373 | case SIOCSIFFLAGS: | |
3374 | case SIOCGIFMETRIC: | |
3375 | case SIOCSIFMETRIC: | |
3376 | case SIOCGIFMTU: | |
3377 | case SIOCSIFMTU: | |
3378 | case SIOCGIFMEM: | |
3379 | case SIOCSIFMEM: | |
3380 | case SIOCGIFHWADDR: | |
3381 | case SIOCSIFHWADDR: | |
3382 | case SIOCADDMULTI: | |
3383 | case SIOCDELMULTI: | |
3384 | case SIOCGIFINDEX: | |
6b96018b AB |
3385 | case SIOCGIFADDR: |
3386 | case SIOCSIFADDR: | |
3387 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3388 | case SIOCDIFADDR: |
6b96018b AB |
3389 | case SIOCGIFBRDADDR: |
3390 | case SIOCSIFBRDADDR: | |
3391 | case SIOCGIFDSTADDR: | |
3392 | case SIOCSIFDSTADDR: | |
3393 | case SIOCGIFNETMASK: | |
3394 | case SIOCSIFNETMASK: | |
3395 | case SIOCSIFPFLAGS: | |
3396 | case SIOCGIFPFLAGS: | |
3397 | case SIOCGIFTXQLEN: | |
3398 | case SIOCSIFTXQLEN: | |
3399 | case SIOCBRADDIF: | |
3400 | case SIOCBRDELIF: | |
c6c9fee3 | 3401 | case SIOCGIFNAME: |
9177efd3 AB |
3402 | case SIOCSIFNAME: |
3403 | case SIOCGMIIPHY: | |
3404 | case SIOCGMIIREG: | |
3405 | case SIOCSMIIREG: | |
f92d4fc9 AV |
3406 | case SIOCBONDENSLAVE: |
3407 | case SIOCBONDRELEASE: | |
3408 | case SIOCBONDSETHWADDR: | |
3409 | case SIOCBONDCHANGEACTIVE: | |
37ac39bd JB |
3410 | return compat_ifreq_ioctl(net, sock, cmd, argp); |
3411 | ||
6b96018b AB |
3412 | case SIOCSARP: |
3413 | case SIOCGARP: | |
3414 | case SIOCDARP: | |
6b96018b | 3415 | case SIOCATMARK: |
63ff03ab | 3416 | return sock_do_ioctl(net, sock, cmd, arg); |
9177efd3 AB |
3417 | } |
3418 | ||
6b96018b AB |
3419 | return -ENOIOCTLCMD; |
3420 | } | |
7a229387 | 3421 | |
95c96174 | 3422 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3423 | unsigned long arg) |
89bbfc95 SP |
3424 | { |
3425 | struct socket *sock = file->private_data; | |
3426 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3427 | struct sock *sk; |
3428 | struct net *net; | |
3429 | ||
3430 | sk = sock->sk; | |
3431 | net = sock_net(sk); | |
89bbfc95 SP |
3432 | |
3433 | if (sock->ops->compat_ioctl) | |
3434 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3435 | ||
87de87d5 DM |
3436 | if (ret == -ENOIOCTLCMD && |
3437 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3438 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3439 | ||
6b96018b AB |
3440 | if (ret == -ENOIOCTLCMD) |
3441 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3442 | ||
89bbfc95 SP |
3443 | return ret; |
3444 | } | |
3445 | #endif | |
3446 | ||
8a3c245c PT |
3447 | /** |
3448 | * kernel_bind - bind an address to a socket (kernel space) | |
3449 | * @sock: socket | |
3450 | * @addr: address | |
3451 | * @addrlen: length of address | |
3452 | * | |
3453 | * Returns 0 or an error. | |
3454 | */ | |
3455 | ||
ac5a488e SS |
3456 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3457 | { | |
3458 | return sock->ops->bind(sock, addr, addrlen); | |
3459 | } | |
c6d409cf | 3460 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e | 3461 | |
8a3c245c PT |
3462 | /** |
3463 | * kernel_listen - move socket to listening state (kernel space) | |
3464 | * @sock: socket | |
3465 | * @backlog: pending connections queue size | |
3466 | * | |
3467 | * Returns 0 or an error. | |
3468 | */ | |
3469 | ||
ac5a488e SS |
3470 | int kernel_listen(struct socket *sock, int backlog) |
3471 | { | |
3472 | return sock->ops->listen(sock, backlog); | |
3473 | } | |
c6d409cf | 3474 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e | 3475 | |
8a3c245c PT |
3476 | /** |
3477 | * kernel_accept - accept a connection (kernel space) | |
3478 | * @sock: listening socket | |
3479 | * @newsock: new connected socket | |
3480 | * @flags: flags | |
3481 | * | |
3482 | * @flags must be SOCK_CLOEXEC, SOCK_NONBLOCK or 0. | |
3483 | * If it fails, @newsock is guaranteed to be %NULL. | |
3484 | * Returns 0 or an error. | |
3485 | */ | |
3486 | ||
ac5a488e SS |
3487 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) |
3488 | { | |
3489 | struct sock *sk = sock->sk; | |
3490 | int err; | |
3491 | ||
3492 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3493 | newsock); | |
3494 | if (err < 0) | |
3495 | goto done; | |
3496 | ||
cdfbabfb | 3497 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3498 | if (err < 0) { |
3499 | sock_release(*newsock); | |
fa8705b0 | 3500 | *newsock = NULL; |
ac5a488e SS |
3501 | goto done; |
3502 | } | |
3503 | ||
3504 | (*newsock)->ops = sock->ops; | |
1b08534e | 3505 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3506 | |
3507 | done: | |
3508 | return err; | |
3509 | } | |
c6d409cf | 3510 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e | 3511 | |
8a3c245c PT |
3512 | /** |
3513 | * kernel_connect - connect a socket (kernel space) | |
3514 | * @sock: socket | |
3515 | * @addr: address | |
3516 | * @addrlen: address length | |
3517 | * @flags: flags (O_NONBLOCK, ...) | |
3518 | * | |
3519 | * For datagram sockets, @addr is the addres to which datagrams are sent | |
3520 | * by default, and the only address from which datagrams are received. | |
3521 | * For stream sockets, attempts to connect to @addr. | |
3522 | * Returns 0 or an error code. | |
3523 | */ | |
3524 | ||
ac5a488e | 3525 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, |
4768fbcb | 3526 | int flags) |
ac5a488e SS |
3527 | { |
3528 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3529 | } | |
c6d409cf | 3530 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e | 3531 | |
8a3c245c PT |
3532 | /** |
3533 | * kernel_getsockname - get the address which the socket is bound (kernel space) | |
3534 | * @sock: socket | |
3535 | * @addr: address holder | |
3536 | * | |
3537 | * Fills the @addr pointer with the address which the socket is bound. | |
3538 | * Returns 0 or an error code. | |
3539 | */ | |
3540 | ||
9b2c45d4 | 3541 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3542 | { |
9b2c45d4 | 3543 | return sock->ops->getname(sock, addr, 0); |
ac5a488e | 3544 | } |
c6d409cf | 3545 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e | 3546 | |
8a3c245c PT |
3547 | /** |
3548 | * kernel_peername - get the address which the socket is connected (kernel space) | |
3549 | * @sock: socket | |
3550 | * @addr: address holder | |
3551 | * | |
3552 | * Fills the @addr pointer with the address which the socket is connected. | |
3553 | * Returns 0 or an error code. | |
3554 | */ | |
3555 | ||
9b2c45d4 | 3556 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3557 | { |
9b2c45d4 | 3558 | return sock->ops->getname(sock, addr, 1); |
ac5a488e | 3559 | } |
c6d409cf | 3560 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e | 3561 | |
8a3c245c PT |
3562 | /** |
3563 | * kernel_getsockopt - get a socket option (kernel space) | |
3564 | * @sock: socket | |
3565 | * @level: API level (SOL_SOCKET, ...) | |
3566 | * @optname: option tag | |
3567 | * @optval: option value | |
3568 | * @optlen: option length | |
3569 | * | |
3570 | * Assigns the option length to @optlen. | |
3571 | * Returns 0 or an error. | |
3572 | */ | |
3573 | ||
ac5a488e SS |
3574 | int kernel_getsockopt(struct socket *sock, int level, int optname, |
3575 | char *optval, int *optlen) | |
3576 | { | |
3577 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3578 | char __user *uoptval; |
3579 | int __user *uoptlen; | |
ac5a488e SS |
3580 | int err; |
3581 | ||
fb8621bb NK |
3582 | uoptval = (char __user __force *) optval; |
3583 | uoptlen = (int __user __force *) optlen; | |
3584 | ||
ac5a488e SS |
3585 | set_fs(KERNEL_DS); |
3586 | if (level == SOL_SOCKET) | |
fb8621bb | 3587 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3588 | else |
fb8621bb NK |
3589 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3590 | uoptlen); | |
ac5a488e SS |
3591 | set_fs(oldfs); |
3592 | return err; | |
3593 | } | |
c6d409cf | 3594 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e | 3595 | |
8a3c245c PT |
3596 | /** |
3597 | * kernel_setsockopt - set a socket option (kernel space) | |
3598 | * @sock: socket | |
3599 | * @level: API level (SOL_SOCKET, ...) | |
3600 | * @optname: option tag | |
3601 | * @optval: option value | |
3602 | * @optlen: option length | |
3603 | * | |
3604 | * Returns 0 or an error. | |
3605 | */ | |
3606 | ||
ac5a488e | 3607 | int kernel_setsockopt(struct socket *sock, int level, int optname, |
b7058842 | 3608 | char *optval, unsigned int optlen) |
ac5a488e SS |
3609 | { |
3610 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3611 | char __user *uoptval; |
ac5a488e SS |
3612 | int err; |
3613 | ||
fb8621bb NK |
3614 | uoptval = (char __user __force *) optval; |
3615 | ||
ac5a488e SS |
3616 | set_fs(KERNEL_DS); |
3617 | if (level == SOL_SOCKET) | |
fb8621bb | 3618 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3619 | else |
fb8621bb | 3620 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3621 | optlen); |
3622 | set_fs(oldfs); | |
3623 | return err; | |
3624 | } | |
c6d409cf | 3625 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e | 3626 | |
8a3c245c PT |
3627 | /** |
3628 | * kernel_sendpage - send a &page through a socket (kernel space) | |
3629 | * @sock: socket | |
3630 | * @page: page | |
3631 | * @offset: page offset | |
3632 | * @size: total size in bytes | |
3633 | * @flags: flags (MSG_DONTWAIT, ...) | |
3634 | * | |
3635 | * Returns the total amount sent in bytes or an error. | |
3636 | */ | |
3637 | ||
ac5a488e SS |
3638 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, |
3639 | size_t size, int flags) | |
3640 | { | |
3641 | if (sock->ops->sendpage) | |
3642 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3643 | ||
3644 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3645 | } | |
c6d409cf | 3646 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3647 | |
8a3c245c PT |
3648 | /** |
3649 | * kernel_sendpage_locked - send a &page through the locked sock (kernel space) | |
3650 | * @sk: sock | |
3651 | * @page: page | |
3652 | * @offset: page offset | |
3653 | * @size: total size in bytes | |
3654 | * @flags: flags (MSG_DONTWAIT, ...) | |
3655 | * | |
3656 | * Returns the total amount sent in bytes or an error. | |
3657 | * Caller must hold @sk. | |
3658 | */ | |
3659 | ||
306b13eb TH |
3660 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3661 | size_t size, int flags) | |
3662 | { | |
3663 | struct socket *sock = sk->sk_socket; | |
3664 | ||
3665 | if (sock->ops->sendpage_locked) | |
3666 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3667 | flags); | |
3668 | ||
3669 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3670 | } | |
3671 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3672 | ||
8a3c245c PT |
3673 | /** |
3674 | * kernel_shutdown - shut down part of a full-duplex connection (kernel space) | |
3675 | * @sock: socket | |
3676 | * @how: connection part | |
3677 | * | |
3678 | * Returns 0 or an error. | |
3679 | */ | |
3680 | ||
91cf45f0 TM |
3681 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3682 | { | |
3683 | return sock->ops->shutdown(sock, how); | |
3684 | } | |
91cf45f0 | 3685 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 | 3686 | |
8a3c245c PT |
3687 | /** |
3688 | * kernel_sock_ip_overhead - returns the IP overhead imposed by a socket | |
3689 | * @sk: socket | |
3690 | * | |
3691 | * This routine returns the IP overhead imposed by a socket i.e. | |
3692 | * the length of the underlying IP header, depending on whether | |
3693 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
3694 | * on at the socket. Assumes that the caller has a lock on the socket. | |
113c3075 | 3695 | */ |
8a3c245c | 3696 | |
113c3075 P |
3697 | u32 kernel_sock_ip_overhead(struct sock *sk) |
3698 | { | |
3699 | struct inet_sock *inet; | |
3700 | struct ip_options_rcu *opt; | |
3701 | u32 overhead = 0; | |
113c3075 P |
3702 | #if IS_ENABLED(CONFIG_IPV6) |
3703 | struct ipv6_pinfo *np; | |
3704 | struct ipv6_txoptions *optv6 = NULL; | |
3705 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3706 | ||
3707 | if (!sk) | |
3708 | return overhead; | |
3709 | ||
113c3075 P |
3710 | switch (sk->sk_family) { |
3711 | case AF_INET: | |
3712 | inet = inet_sk(sk); | |
3713 | overhead += sizeof(struct iphdr); | |
3714 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3715 | sock_owned_by_user(sk)); |
113c3075 P |
3716 | if (opt) |
3717 | overhead += opt->opt.optlen; | |
3718 | return overhead; | |
3719 | #if IS_ENABLED(CONFIG_IPV6) | |
3720 | case AF_INET6: | |
3721 | np = inet6_sk(sk); | |
3722 | overhead += sizeof(struct ipv6hdr); | |
3723 | if (np) | |
3724 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3725 | sock_owned_by_user(sk)); |
113c3075 P |
3726 | if (optv6) |
3727 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3728 | return overhead; | |
3729 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3730 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3731 | return overhead; | |
3732 | } | |
3733 | } | |
3734 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |