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 | 75 | #include <linux/mount.h> |
fba9be49 | 76 | #include <linux/pseudo_fs.h> |
1da177e4 LT |
77 | #include <linux/security.h> |
78 | #include <linux/syscalls.h> | |
79 | #include <linux/compat.h> | |
80 | #include <linux/kmod.h> | |
3ec3b2fb | 81 | #include <linux/audit.h> |
d86b5e0e | 82 | #include <linux/wireless.h> |
1b8d7ae4 | 83 | #include <linux/nsproxy.h> |
1fd7317d | 84 | #include <linux/magic.h> |
5a0e3ad6 | 85 | #include <linux/slab.h> |
600e1779 | 86 | #include <linux/xattr.h> |
c8e8cd57 | 87 | #include <linux/nospec.h> |
8c3c447b | 88 | #include <linux/indirect_call_wrapper.h> |
1da177e4 | 89 | |
7c0f6ba6 | 90 | #include <linux/uaccess.h> |
1da177e4 LT |
91 | #include <asm/unistd.h> |
92 | ||
93 | #include <net/compat.h> | |
87de87d5 | 94 | #include <net/wext.h> |
f8451725 | 95 | #include <net/cls_cgroup.h> |
1da177e4 LT |
96 | |
97 | #include <net/sock.h> | |
98 | #include <linux/netfilter.h> | |
99 | ||
6b96018b AB |
100 | #include <linux/if_tun.h> |
101 | #include <linux/ipv6_route.h> | |
102 | #include <linux/route.h> | |
6b96018b | 103 | #include <linux/sockios.h> |
076bb0c8 | 104 | #include <net/busy_poll.h> |
f24b9be5 | 105 | #include <linux/errqueue.h> |
06021292 | 106 | |
e0d1095a | 107 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
108 | unsigned int sysctl_net_busy_read __read_mostly; |
109 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 110 | #endif |
6b96018b | 111 | |
8ae5e030 AV |
112 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
113 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 114 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
115 | |
116 | static int sock_close(struct inode *inode, struct file *file); | |
a11e1d43 LT |
117 | static __poll_t sock_poll(struct file *file, |
118 | struct poll_table_struct *wait); | |
89bddce5 | 119 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
120 | #ifdef CONFIG_COMPAT |
121 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 122 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 123 | #endif |
1da177e4 | 124 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
125 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
126 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 127 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 128 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 129 | unsigned int flags); |
1da177e4 | 130 | |
1da177e4 LT |
131 | /* |
132 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
133 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
134 | */ | |
135 | ||
da7071d7 | 136 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
137 | .owner = THIS_MODULE, |
138 | .llseek = no_llseek, | |
8ae5e030 AV |
139 | .read_iter = sock_read_iter, |
140 | .write_iter = sock_write_iter, | |
1da177e4 LT |
141 | .poll = sock_poll, |
142 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
143 | #ifdef CONFIG_COMPAT |
144 | .compat_ioctl = compat_sock_ioctl, | |
145 | #endif | |
1da177e4 | 146 | .mmap = sock_mmap, |
1da177e4 LT |
147 | .release = sock_close, |
148 | .fasync = sock_fasync, | |
5274f052 JA |
149 | .sendpage = sock_sendpage, |
150 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 151 | .splice_read = sock_splice_read, |
1da177e4 LT |
152 | }; |
153 | ||
154 | /* | |
155 | * The protocol list. Each protocol is registered in here. | |
156 | */ | |
157 | ||
1da177e4 | 158 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 159 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 160 | |
1da177e4 | 161 | /* |
89bddce5 SH |
162 | * Support routines. |
163 | * Move socket addresses back and forth across the kernel/user | |
164 | * divide and look after the messy bits. | |
1da177e4 LT |
165 | */ |
166 | ||
1da177e4 LT |
167 | /** |
168 | * move_addr_to_kernel - copy a socket address into kernel space | |
169 | * @uaddr: Address in user space | |
170 | * @kaddr: Address in kernel space | |
171 | * @ulen: Length in user space | |
172 | * | |
173 | * The address is copied into kernel space. If the provided address is | |
174 | * too long an error code of -EINVAL is returned. If the copy gives | |
175 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
176 | */ | |
177 | ||
43db362d | 178 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 179 | { |
230b1839 | 180 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 181 | return -EINVAL; |
89bddce5 | 182 | if (ulen == 0) |
1da177e4 | 183 | return 0; |
89bddce5 | 184 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 185 | return -EFAULT; |
3ec3b2fb | 186 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
187 | } |
188 | ||
189 | /** | |
190 | * move_addr_to_user - copy an address to user space | |
191 | * @kaddr: kernel space address | |
192 | * @klen: length of address in kernel | |
193 | * @uaddr: user space address | |
194 | * @ulen: pointer to user length field | |
195 | * | |
196 | * The value pointed to by ulen on entry is the buffer length available. | |
197 | * This is overwritten with the buffer space used. -EINVAL is returned | |
198 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
199 | * is returned if either the buffer or the length field are not | |
200 | * accessible. | |
201 | * After copying the data up to the limit the user specifies, the true | |
202 | * length of the data is written over the length limit the user | |
203 | * specified. Zero is returned for a success. | |
204 | */ | |
89bddce5 | 205 | |
43db362d | 206 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 207 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
208 | { |
209 | int err; | |
210 | int len; | |
211 | ||
68c6beb3 | 212 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
213 | err = get_user(len, ulen); |
214 | if (err) | |
1da177e4 | 215 | return err; |
89bddce5 SH |
216 | if (len > klen) |
217 | len = klen; | |
68c6beb3 | 218 | if (len < 0) |
1da177e4 | 219 | return -EINVAL; |
89bddce5 | 220 | if (len) { |
d6fe3945 SG |
221 | if (audit_sockaddr(klen, kaddr)) |
222 | return -ENOMEM; | |
89bddce5 | 223 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
224 | return -EFAULT; |
225 | } | |
226 | /* | |
89bddce5 SH |
227 | * "fromlen shall refer to the value before truncation.." |
228 | * 1003.1g | |
1da177e4 LT |
229 | */ |
230 | return __put_user(klen, ulen); | |
231 | } | |
232 | ||
08009a76 | 233 | static struct kmem_cache *sock_inode_cachep __ro_after_init; |
1da177e4 LT |
234 | |
235 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
236 | { | |
237 | struct socket_alloc *ei; | |
89bddce5 | 238 | |
e94b1766 | 239 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
240 | if (!ei) |
241 | return NULL; | |
333f7909 AV |
242 | init_waitqueue_head(&ei->socket.wq.wait); |
243 | ei->socket.wq.fasync_list = NULL; | |
244 | ei->socket.wq.flags = 0; | |
89bddce5 | 245 | |
1da177e4 LT |
246 | ei->socket.state = SS_UNCONNECTED; |
247 | ei->socket.flags = 0; | |
248 | ei->socket.ops = NULL; | |
249 | ei->socket.sk = NULL; | |
250 | ei->socket.file = NULL; | |
1da177e4 LT |
251 | |
252 | return &ei->vfs_inode; | |
253 | } | |
254 | ||
6d7855c5 | 255 | static void sock_free_inode(struct inode *inode) |
1da177e4 | 256 | { |
43815482 ED |
257 | struct socket_alloc *ei; |
258 | ||
259 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
43815482 | 260 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
261 | } |
262 | ||
51cc5068 | 263 | static void init_once(void *foo) |
1da177e4 | 264 | { |
89bddce5 | 265 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 266 | |
a35afb83 | 267 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 268 | } |
89bddce5 | 269 | |
1e911632 | 270 | static void init_inodecache(void) |
1da177e4 LT |
271 | { |
272 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
273 | sizeof(struct socket_alloc), |
274 | 0, | |
275 | (SLAB_HWCACHE_ALIGN | | |
276 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 277 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 278 | init_once); |
1e911632 | 279 | BUG_ON(sock_inode_cachep == NULL); |
1da177e4 LT |
280 | } |
281 | ||
b87221de | 282 | static const struct super_operations sockfs_ops = { |
c6d409cf | 283 | .alloc_inode = sock_alloc_inode, |
6d7855c5 | 284 | .free_inode = sock_free_inode, |
c6d409cf | 285 | .statfs = simple_statfs, |
1da177e4 LT |
286 | }; |
287 | ||
c23fbb6b ED |
288 | /* |
289 | * sockfs_dname() is called from d_path(). | |
290 | */ | |
291 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
292 | { | |
293 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 294 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
295 | } |
296 | ||
3ba13d17 | 297 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 298 | .d_dname = sockfs_dname, |
1da177e4 LT |
299 | }; |
300 | ||
bba0bd31 AG |
301 | static int sockfs_xattr_get(const struct xattr_handler *handler, |
302 | struct dentry *dentry, struct inode *inode, | |
303 | const char *suffix, void *value, size_t size) | |
304 | { | |
305 | if (value) { | |
306 | if (dentry->d_name.len + 1 > size) | |
307 | return -ERANGE; | |
308 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
309 | } | |
310 | return dentry->d_name.len + 1; | |
311 | } | |
312 | ||
313 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" | |
314 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
315 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
316 | ||
317 | static const struct xattr_handler sockfs_xattr_handler = { | |
318 | .name = XATTR_NAME_SOCKPROTONAME, | |
319 | .get = sockfs_xattr_get, | |
320 | }; | |
321 | ||
4a590153 AG |
322 | static int sockfs_security_xattr_set(const struct xattr_handler *handler, |
323 | struct dentry *dentry, struct inode *inode, | |
324 | const char *suffix, const void *value, | |
325 | size_t size, int flags) | |
326 | { | |
327 | /* Handled by LSM. */ | |
328 | return -EAGAIN; | |
329 | } | |
330 | ||
331 | static const struct xattr_handler sockfs_security_xattr_handler = { | |
332 | .prefix = XATTR_SECURITY_PREFIX, | |
333 | .set = sockfs_security_xattr_set, | |
334 | }; | |
335 | ||
bba0bd31 AG |
336 | static const struct xattr_handler *sockfs_xattr_handlers[] = { |
337 | &sockfs_xattr_handler, | |
4a590153 | 338 | &sockfs_security_xattr_handler, |
bba0bd31 AG |
339 | NULL |
340 | }; | |
341 | ||
fba9be49 | 342 | static int sockfs_init_fs_context(struct fs_context *fc) |
c74a1cbb | 343 | { |
fba9be49 DH |
344 | struct pseudo_fs_context *ctx = init_pseudo(fc, SOCKFS_MAGIC); |
345 | if (!ctx) | |
346 | return -ENOMEM; | |
347 | ctx->ops = &sockfs_ops; | |
348 | ctx->dops = &sockfs_dentry_operations; | |
349 | ctx->xattr = sockfs_xattr_handlers; | |
350 | return 0; | |
c74a1cbb AV |
351 | } |
352 | ||
353 | static struct vfsmount *sock_mnt __read_mostly; | |
354 | ||
355 | static struct file_system_type sock_fs_type = { | |
356 | .name = "sockfs", | |
fba9be49 | 357 | .init_fs_context = sockfs_init_fs_context, |
c74a1cbb AV |
358 | .kill_sb = kill_anon_super, |
359 | }; | |
360 | ||
1da177e4 LT |
361 | /* |
362 | * Obtains the first available file descriptor and sets it up for use. | |
363 | * | |
39d8c1b6 DM |
364 | * These functions create file structures and maps them to fd space |
365 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
366 | * and file struct implicitly stored in sock->file. |
367 | * Note that another thread may close file descriptor before we return | |
368 | * from this function. We use the fact that now we do not refer | |
369 | * to socket after mapping. If one day we will need it, this | |
370 | * function will increment ref. count on file by 1. | |
371 | * | |
372 | * In any case returned fd MAY BE not valid! | |
373 | * This race condition is unavoidable | |
374 | * with shared fd spaces, we cannot solve it inside kernel, | |
375 | * but we take care of internal coherence yet. | |
376 | */ | |
377 | ||
8a3c245c PT |
378 | /** |
379 | * sock_alloc_file - Bind a &socket to a &file | |
380 | * @sock: socket | |
381 | * @flags: file status flags | |
382 | * @dname: protocol name | |
383 | * | |
384 | * Returns the &file bound with @sock, implicitly storing it | |
385 | * in sock->file. If dname is %NULL, sets to "". | |
386 | * On failure the return is a ERR pointer (see linux/err.h). | |
387 | * This function uses GFP_KERNEL internally. | |
388 | */ | |
389 | ||
aab174f0 | 390 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 391 | { |
7cbe66b6 | 392 | struct file *file; |
1da177e4 | 393 | |
d93aa9d8 AV |
394 | if (!dname) |
395 | dname = sock->sk ? sock->sk->sk_prot_creator->name : ""; | |
39d8c1b6 | 396 | |
d93aa9d8 AV |
397 | file = alloc_file_pseudo(SOCK_INODE(sock), sock_mnt, dname, |
398 | O_RDWR | (flags & O_NONBLOCK), | |
399 | &socket_file_ops); | |
b5ffe634 | 400 | if (IS_ERR(file)) { |
8e1611e2 | 401 | sock_release(sock); |
39b65252 | 402 | return file; |
cc3808f8 AV |
403 | } |
404 | ||
405 | sock->file = file; | |
39d8c1b6 | 406 | file->private_data = sock; |
d8e464ec | 407 | stream_open(SOCK_INODE(sock), file); |
28407630 | 408 | return file; |
39d8c1b6 | 409 | } |
56b31d1c | 410 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 411 | |
56b31d1c | 412 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
413 | { |
414 | struct file *newfile; | |
28407630 | 415 | int fd = get_unused_fd_flags(flags); |
ce4bb04c AV |
416 | if (unlikely(fd < 0)) { |
417 | sock_release(sock); | |
28407630 | 418 | return fd; |
ce4bb04c | 419 | } |
39d8c1b6 | 420 | |
aab174f0 | 421 | newfile = sock_alloc_file(sock, flags, NULL); |
4546e44c | 422 | if (!IS_ERR(newfile)) { |
39d8c1b6 | 423 | fd_install(fd, newfile); |
28407630 AV |
424 | return fd; |
425 | } | |
7cbe66b6 | 426 | |
28407630 AV |
427 | put_unused_fd(fd); |
428 | return PTR_ERR(newfile); | |
1da177e4 LT |
429 | } |
430 | ||
8a3c245c PT |
431 | /** |
432 | * sock_from_file - Return the &socket bounded to @file. | |
433 | * @file: file | |
434 | * @err: pointer to an error code return | |
435 | * | |
436 | * On failure returns %NULL and assigns -ENOTSOCK to @err. | |
437 | */ | |
438 | ||
406a3c63 | 439 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 440 | { |
6cb153ca BL |
441 | if (file->f_op == &socket_file_ops) |
442 | return file->private_data; /* set in sock_map_fd */ | |
443 | ||
23bb80d2 ED |
444 | *err = -ENOTSOCK; |
445 | return NULL; | |
6cb153ca | 446 | } |
406a3c63 | 447 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 448 | |
1da177e4 | 449 | /** |
c6d409cf | 450 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
451 | * @fd: file handle |
452 | * @err: pointer to an error code return | |
453 | * | |
454 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 455 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
456 | * with a negative errno code and NULL is returned. The function checks |
457 | * for both invalid handles and passing a handle which is not a socket. | |
458 | * | |
459 | * On a success the socket object pointer is returned. | |
460 | */ | |
461 | ||
462 | struct socket *sockfd_lookup(int fd, int *err) | |
463 | { | |
464 | struct file *file; | |
1da177e4 LT |
465 | struct socket *sock; |
466 | ||
89bddce5 SH |
467 | file = fget(fd); |
468 | if (!file) { | |
1da177e4 LT |
469 | *err = -EBADF; |
470 | return NULL; | |
471 | } | |
89bddce5 | 472 | |
6cb153ca BL |
473 | sock = sock_from_file(file, err); |
474 | if (!sock) | |
1da177e4 | 475 | fput(file); |
6cb153ca BL |
476 | return sock; |
477 | } | |
c6d409cf | 478 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 479 | |
6cb153ca BL |
480 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
481 | { | |
00e188ef | 482 | struct fd f = fdget(fd); |
6cb153ca BL |
483 | struct socket *sock; |
484 | ||
3672558c | 485 | *err = -EBADF; |
00e188ef AV |
486 | if (f.file) { |
487 | sock = sock_from_file(f.file, err); | |
488 | if (likely(sock)) { | |
489 | *fput_needed = f.flags; | |
6cb153ca | 490 | return sock; |
00e188ef AV |
491 | } |
492 | fdput(f); | |
1da177e4 | 493 | } |
6cb153ca | 494 | return NULL; |
1da177e4 LT |
495 | } |
496 | ||
600e1779 MY |
497 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
498 | size_t size) | |
499 | { | |
500 | ssize_t len; | |
501 | ssize_t used = 0; | |
502 | ||
c5ef6035 | 503 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
504 | if (len < 0) |
505 | return len; | |
506 | used += len; | |
507 | if (buffer) { | |
508 | if (size < used) | |
509 | return -ERANGE; | |
510 | buffer += len; | |
511 | } | |
512 | ||
513 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
514 | used += len; | |
515 | if (buffer) { | |
516 | if (size < used) | |
517 | return -ERANGE; | |
518 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
519 | buffer += len; | |
520 | } | |
521 | ||
522 | return used; | |
523 | } | |
524 | ||
dc647ec8 | 525 | static int sockfs_setattr(struct dentry *dentry, struct iattr *iattr) |
86741ec2 LC |
526 | { |
527 | int err = simple_setattr(dentry, iattr); | |
528 | ||
e1a3a60a | 529 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
530 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
531 | ||
6d8c50dc CW |
532 | if (sock->sk) |
533 | sock->sk->sk_uid = iattr->ia_uid; | |
534 | else | |
535 | err = -ENOENT; | |
86741ec2 LC |
536 | } |
537 | ||
538 | return err; | |
539 | } | |
540 | ||
600e1779 | 541 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 542 | .listxattr = sockfs_listxattr, |
86741ec2 | 543 | .setattr = sockfs_setattr, |
600e1779 MY |
544 | }; |
545 | ||
1da177e4 | 546 | /** |
8a3c245c | 547 | * sock_alloc - allocate a socket |
89bddce5 | 548 | * |
1da177e4 LT |
549 | * Allocate a new inode and socket object. The two are bound together |
550 | * and initialised. The socket is then returned. If we are out of inodes | |
8a3c245c | 551 | * NULL is returned. This functions uses GFP_KERNEL internally. |
1da177e4 LT |
552 | */ |
553 | ||
f4a00aac | 554 | struct socket *sock_alloc(void) |
1da177e4 | 555 | { |
89bddce5 SH |
556 | struct inode *inode; |
557 | struct socket *sock; | |
1da177e4 | 558 | |
a209dfc7 | 559 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
560 | if (!inode) |
561 | return NULL; | |
562 | ||
563 | sock = SOCKET_I(inode); | |
564 | ||
85fe4025 | 565 | inode->i_ino = get_next_ino(); |
89bddce5 | 566 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
567 | inode->i_uid = current_fsuid(); |
568 | inode->i_gid = current_fsgid(); | |
600e1779 | 569 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 570 | |
1da177e4 LT |
571 | return sock; |
572 | } | |
f4a00aac | 573 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 574 | |
1da177e4 | 575 | /** |
8a3c245c | 576 | * sock_release - close a socket |
1da177e4 LT |
577 | * @sock: socket to close |
578 | * | |
579 | * The socket is released from the protocol stack if it has a release | |
580 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 581 | * an inode not a file. |
1da177e4 | 582 | */ |
89bddce5 | 583 | |
6d8c50dc | 584 | static void __sock_release(struct socket *sock, struct inode *inode) |
1da177e4 LT |
585 | { |
586 | if (sock->ops) { | |
587 | struct module *owner = sock->ops->owner; | |
588 | ||
6d8c50dc CW |
589 | if (inode) |
590 | inode_lock(inode); | |
1da177e4 | 591 | sock->ops->release(sock); |
ff7b11aa | 592 | sock->sk = NULL; |
6d8c50dc CW |
593 | if (inode) |
594 | inode_unlock(inode); | |
1da177e4 LT |
595 | sock->ops = NULL; |
596 | module_put(owner); | |
597 | } | |
598 | ||
333f7909 | 599 | if (sock->wq.fasync_list) |
3410f22e | 600 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 601 | |
1da177e4 LT |
602 | if (!sock->file) { |
603 | iput(SOCK_INODE(sock)); | |
604 | return; | |
605 | } | |
89bddce5 | 606 | sock->file = NULL; |
1da177e4 | 607 | } |
6d8c50dc CW |
608 | |
609 | void sock_release(struct socket *sock) | |
610 | { | |
611 | __sock_release(sock, NULL); | |
612 | } | |
c6d409cf | 613 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 614 | |
c14ac945 | 615 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 616 | { |
140c55d4 ED |
617 | u8 flags = *tx_flags; |
618 | ||
c14ac945 | 619 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
620 | flags |= SKBTX_HW_TSTAMP; |
621 | ||
c14ac945 | 622 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
623 | flags |= SKBTX_SW_TSTAMP; |
624 | ||
c14ac945 | 625 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
626 | flags |= SKBTX_SCHED_TSTAMP; |
627 | ||
140c55d4 | 628 | *tx_flags = flags; |
20d49473 | 629 | } |
67cc0d40 | 630 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 631 | |
8c3c447b PA |
632 | INDIRECT_CALLABLE_DECLARE(int inet_sendmsg(struct socket *, struct msghdr *, |
633 | size_t)); | |
a648a592 PA |
634 | INDIRECT_CALLABLE_DECLARE(int inet6_sendmsg(struct socket *, struct msghdr *, |
635 | size_t)); | |
d8725c86 | 636 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 637 | { |
a648a592 PA |
638 | int ret = INDIRECT_CALL_INET(sock->ops->sendmsg, inet6_sendmsg, |
639 | inet_sendmsg, sock, msg, | |
640 | msg_data_left(msg)); | |
d8725c86 AV |
641 | BUG_ON(ret == -EIOCBQUEUED); |
642 | return ret; | |
1da177e4 LT |
643 | } |
644 | ||
85806af0 RD |
645 | /** |
646 | * sock_sendmsg - send a message through @sock | |
647 | * @sock: socket | |
648 | * @msg: message to send | |
649 | * | |
650 | * Sends @msg through @sock, passing through LSM. | |
651 | * Returns the number of bytes sent, or an error code. | |
652 | */ | |
d8725c86 | 653 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 654 | { |
d8725c86 | 655 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 656 | msg_data_left(msg)); |
228e548e | 657 | |
d8725c86 | 658 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 659 | } |
c6d409cf | 660 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 | 661 | |
8a3c245c PT |
662 | /** |
663 | * kernel_sendmsg - send a message through @sock (kernel-space) | |
664 | * @sock: socket | |
665 | * @msg: message header | |
666 | * @vec: kernel vec | |
667 | * @num: vec array length | |
668 | * @size: total message data size | |
669 | * | |
670 | * Builds the message data with @vec and sends it through @sock. | |
671 | * Returns the number of bytes sent, or an error code. | |
672 | */ | |
673 | ||
1da177e4 LT |
674 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
675 | struct kvec *vec, size_t num, size_t size) | |
676 | { | |
aa563d7b | 677 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
d8725c86 | 678 | return sock_sendmsg(sock, msg); |
1da177e4 | 679 | } |
c6d409cf | 680 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 681 | |
8a3c245c PT |
682 | /** |
683 | * kernel_sendmsg_locked - send a message through @sock (kernel-space) | |
684 | * @sk: sock | |
685 | * @msg: message header | |
686 | * @vec: output s/g array | |
687 | * @num: output s/g array length | |
688 | * @size: total message data size | |
689 | * | |
690 | * Builds the message data with @vec and sends it through @sock. | |
691 | * Returns the number of bytes sent, or an error code. | |
692 | * Caller must hold @sk. | |
693 | */ | |
694 | ||
306b13eb TH |
695 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
696 | struct kvec *vec, size_t num, size_t size) | |
697 | { | |
698 | struct socket *sock = sk->sk_socket; | |
699 | ||
700 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 701 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb | 702 | |
aa563d7b | 703 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
306b13eb TH |
704 | |
705 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
706 | } | |
707 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
708 | ||
8605330a SHY |
709 | static bool skb_is_err_queue(const struct sk_buff *skb) |
710 | { | |
711 | /* pkt_type of skbs enqueued on the error queue are set to | |
712 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
713 | * in recvmsg, since skbs received on a local socket will never | |
714 | * have a pkt_type of PACKET_OUTGOING. | |
715 | */ | |
716 | return skb->pkt_type == PACKET_OUTGOING; | |
717 | } | |
718 | ||
b50a5c70 ML |
719 | /* On transmit, software and hardware timestamps are returned independently. |
720 | * As the two skb clones share the hardware timestamp, which may be updated | |
721 | * before the software timestamp is received, a hardware TX timestamp may be | |
722 | * returned only if there is no software TX timestamp. Ignore false software | |
723 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
7f1bc6e9 | 724 | * option SO_TIMESTAMP_OLD(NS) is enabled on the socket, even when the skb has a |
b50a5c70 ML |
725 | * hardware timestamp. |
726 | */ | |
727 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
728 | { | |
729 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
730 | } | |
731 | ||
aad9c8c4 ML |
732 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb) |
733 | { | |
734 | struct scm_ts_pktinfo ts_pktinfo; | |
735 | struct net_device *orig_dev; | |
736 | ||
737 | if (!skb_mac_header_was_set(skb)) | |
738 | return; | |
739 | ||
740 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
741 | ||
742 | rcu_read_lock(); | |
743 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
744 | if (orig_dev) | |
745 | ts_pktinfo.if_index = orig_dev->ifindex; | |
746 | rcu_read_unlock(); | |
747 | ||
748 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
749 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
750 | sizeof(ts_pktinfo), &ts_pktinfo); | |
751 | } | |
752 | ||
92f37fd2 ED |
753 | /* |
754 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
755 | */ | |
756 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
757 | struct sk_buff *skb) | |
758 | { | |
20d49473 | 759 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
887feae3 | 760 | int new_tstamp = sock_flag(sk, SOCK_TSTAMP_NEW); |
9718475e DD |
761 | struct scm_timestamping_internal tss; |
762 | ||
b50a5c70 | 763 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
764 | struct skb_shared_hwtstamps *shhwtstamps = |
765 | skb_hwtstamps(skb); | |
766 | ||
767 | /* Race occurred between timestamp enabling and packet | |
768 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 769 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 770 | __net_timestamp(skb); |
b50a5c70 ML |
771 | false_tstamp = 1; |
772 | } | |
20d49473 PO |
773 | |
774 | if (need_software_tstamp) { | |
775 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
887feae3 DD |
776 | if (new_tstamp) { |
777 | struct __kernel_sock_timeval tv; | |
778 | ||
779 | skb_get_new_timestamp(skb, &tv); | |
780 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_NEW, | |
781 | sizeof(tv), &tv); | |
782 | } else { | |
783 | struct __kernel_old_timeval tv; | |
784 | ||
785 | skb_get_timestamp(skb, &tv); | |
786 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_OLD, | |
787 | sizeof(tv), &tv); | |
788 | } | |
20d49473 | 789 | } else { |
887feae3 DD |
790 | if (new_tstamp) { |
791 | struct __kernel_timespec ts; | |
792 | ||
793 | skb_get_new_timestampns(skb, &ts); | |
794 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_NEW, | |
795 | sizeof(ts), &ts); | |
796 | } else { | |
797 | struct timespec ts; | |
798 | ||
799 | skb_get_timestampns(skb, &ts); | |
800 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_OLD, | |
801 | sizeof(ts), &ts); | |
802 | } | |
20d49473 PO |
803 | } |
804 | } | |
805 | ||
f24b9be5 | 806 | memset(&tss, 0, sizeof(tss)); |
c199105d | 807 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
9718475e | 808 | ktime_to_timespec64_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 809 | empty = 0; |
4d276eb6 | 810 | if (shhwtstamps && |
b9f40e21 | 811 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
b50a5c70 | 812 | !skb_is_swtx_tstamp(skb, false_tstamp) && |
9718475e | 813 | ktime_to_timespec64_cond(shhwtstamps->hwtstamp, tss.ts + 2)) { |
4d276eb6 | 814 | empty = 0; |
aad9c8c4 ML |
815 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && |
816 | !skb_is_err_queue(skb)) | |
817 | put_ts_pktinfo(msg, skb); | |
818 | } | |
1c885808 | 819 | if (!empty) { |
9718475e DD |
820 | if (sock_flag(sk, SOCK_TSTAMP_NEW)) |
821 | put_cmsg_scm_timestamping64(msg, &tss); | |
822 | else | |
823 | put_cmsg_scm_timestamping(msg, &tss); | |
1c885808 | 824 | |
8605330a | 825 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 826 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
827 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
828 | skb->len, skb->data); | |
829 | } | |
92f37fd2 | 830 | } |
7c81fd8b ACM |
831 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
832 | ||
6e3e939f JB |
833 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
834 | struct sk_buff *skb) | |
835 | { | |
836 | int ack; | |
837 | ||
838 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
839 | return; | |
840 | if (!skb->wifi_acked_valid) | |
841 | return; | |
842 | ||
843 | ack = skb->wifi_acked; | |
844 | ||
845 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
846 | } | |
847 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
848 | ||
11165f14 | 849 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
850 | struct sk_buff *skb) | |
3b885787 | 851 | { |
744d5a3e | 852 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 853 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 854 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
855 | } |
856 | ||
767dd033 | 857 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
858 | struct sk_buff *skb) |
859 | { | |
860 | sock_recv_timestamp(msg, sk, skb); | |
861 | sock_recv_drops(msg, sk, skb); | |
862 | } | |
767dd033 | 863 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 864 | |
8c3c447b | 865 | INDIRECT_CALLABLE_DECLARE(int inet_recvmsg(struct socket *, struct msghdr *, |
a648a592 PA |
866 | size_t, int)); |
867 | INDIRECT_CALLABLE_DECLARE(int inet6_recvmsg(struct socket *, struct msghdr *, | |
868 | size_t, int)); | |
1b784140 | 869 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 870 | int flags) |
1da177e4 | 871 | { |
a648a592 PA |
872 | return INDIRECT_CALL_INET(sock->ops->recvmsg, inet6_recvmsg, |
873 | inet_recvmsg, sock, msg, msg_data_left(msg), | |
874 | flags); | |
1da177e4 LT |
875 | } |
876 | ||
85806af0 RD |
877 | /** |
878 | * sock_recvmsg - receive a message from @sock | |
879 | * @sock: socket | |
880 | * @msg: message to receive | |
881 | * @flags: message flags | |
882 | * | |
883 | * Receives @msg from @sock, passing through LSM. Returns the total number | |
884 | * of bytes received, or an error. | |
885 | */ | |
2da62906 | 886 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 887 | { |
2da62906 | 888 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 889 | |
2da62906 | 890 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 891 | } |
c6d409cf | 892 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 893 | |
c1249c0a | 894 | /** |
8a3c245c PT |
895 | * kernel_recvmsg - Receive a message from a socket (kernel space) |
896 | * @sock: The socket to receive the message from | |
897 | * @msg: Received message | |
898 | * @vec: Input s/g array for message data | |
899 | * @num: Size of input s/g array | |
900 | * @size: Number of bytes to read | |
901 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
c1249c0a | 902 | * |
8a3c245c PT |
903 | * On return the msg structure contains the scatter/gather array passed in the |
904 | * vec argument. The array is modified so that it consists of the unfilled | |
905 | * portion of the original array. | |
c1249c0a | 906 | * |
8a3c245c | 907 | * The returned value is the total number of bytes received, or an error. |
c1249c0a | 908 | */ |
8a3c245c | 909 | |
89bddce5 SH |
910 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
911 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
912 | { |
913 | mm_segment_t oldfs = get_fs(); | |
914 | int result; | |
915 | ||
aa563d7b | 916 | iov_iter_kvec(&msg->msg_iter, READ, vec, num, size); |
1da177e4 | 917 | set_fs(KERNEL_DS); |
2da62906 | 918 | result = sock_recvmsg(sock, msg, flags); |
1da177e4 LT |
919 | set_fs(oldfs); |
920 | return result; | |
921 | } | |
c6d409cf | 922 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 923 | |
ce1d4d3e CH |
924 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
925 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 926 | { |
1da177e4 LT |
927 | struct socket *sock; |
928 | int flags; | |
929 | ||
ce1d4d3e CH |
930 | sock = file->private_data; |
931 | ||
35f9c09f ED |
932 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
933 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
934 | flags |= more; | |
ce1d4d3e | 935 | |
e6949583 | 936 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 937 | } |
1da177e4 | 938 | |
9c55e01c | 939 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 940 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
941 | unsigned int flags) |
942 | { | |
943 | struct socket *sock = file->private_data; | |
944 | ||
997b37da | 945 | if (unlikely(!sock->ops->splice_read)) |
95506588 | 946 | return generic_file_splice_read(file, ppos, pipe, len, flags); |
997b37da | 947 | |
9c55e01c JA |
948 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
949 | } | |
950 | ||
8ae5e030 | 951 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 952 | { |
6d652330 AV |
953 | struct file *file = iocb->ki_filp; |
954 | struct socket *sock = file->private_data; | |
0345f931 | 955 | struct msghdr msg = {.msg_iter = *to, |
956 | .msg_iocb = iocb}; | |
8ae5e030 | 957 | ssize_t res; |
ce1d4d3e | 958 | |
8ae5e030 AV |
959 | if (file->f_flags & O_NONBLOCK) |
960 | msg.msg_flags = MSG_DONTWAIT; | |
961 | ||
962 | if (iocb->ki_pos != 0) | |
1da177e4 | 963 | return -ESPIPE; |
027445c3 | 964 | |
66ee59af | 965 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
966 | return 0; |
967 | ||
2da62906 | 968 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
969 | *to = msg.msg_iter; |
970 | return res; | |
1da177e4 LT |
971 | } |
972 | ||
8ae5e030 | 973 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 974 | { |
6d652330 AV |
975 | struct file *file = iocb->ki_filp; |
976 | struct socket *sock = file->private_data; | |
0345f931 | 977 | struct msghdr msg = {.msg_iter = *from, |
978 | .msg_iocb = iocb}; | |
8ae5e030 | 979 | ssize_t res; |
1da177e4 | 980 | |
8ae5e030 | 981 | if (iocb->ki_pos != 0) |
ce1d4d3e | 982 | return -ESPIPE; |
027445c3 | 983 | |
8ae5e030 AV |
984 | if (file->f_flags & O_NONBLOCK) |
985 | msg.msg_flags = MSG_DONTWAIT; | |
986 | ||
6d652330 AV |
987 | if (sock->type == SOCK_SEQPACKET) |
988 | msg.msg_flags |= MSG_EOR; | |
989 | ||
d8725c86 | 990 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
991 | *from = msg.msg_iter; |
992 | return res; | |
1da177e4 LT |
993 | } |
994 | ||
1da177e4 LT |
995 | /* |
996 | * Atomic setting of ioctl hooks to avoid race | |
997 | * with module unload. | |
998 | */ | |
999 | ||
4a3e2f71 | 1000 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 1001 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 1002 | |
881d966b | 1003 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 1004 | { |
4a3e2f71 | 1005 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 1006 | br_ioctl_hook = hook; |
4a3e2f71 | 1007 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1008 | } |
1009 | EXPORT_SYMBOL(brioctl_set); | |
1010 | ||
4a3e2f71 | 1011 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 1012 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 1013 | |
881d966b | 1014 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 1015 | { |
4a3e2f71 | 1016 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1017 | vlan_ioctl_hook = hook; |
4a3e2f71 | 1018 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
1019 | } |
1020 | EXPORT_SYMBOL(vlan_ioctl_set); | |
1021 | ||
4a3e2f71 | 1022 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 1023 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 1024 | |
89bddce5 | 1025 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 1026 | { |
4a3e2f71 | 1027 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 1028 | dlci_ioctl_hook = hook; |
4a3e2f71 | 1029 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
1030 | } |
1031 | EXPORT_SYMBOL(dlci_ioctl_set); | |
1032 | ||
6b96018b | 1033 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
63ff03ab | 1034 | unsigned int cmd, unsigned long arg) |
6b96018b AB |
1035 | { |
1036 | int err; | |
1037 | void __user *argp = (void __user *)arg; | |
1038 | ||
1039 | err = sock->ops->ioctl(sock, cmd, arg); | |
1040 | ||
1041 | /* | |
1042 | * If this ioctl is unknown try to hand it down | |
1043 | * to the NIC driver. | |
1044 | */ | |
36fd633e AV |
1045 | if (err != -ENOIOCTLCMD) |
1046 | return err; | |
6b96018b | 1047 | |
36fd633e AV |
1048 | if (cmd == SIOCGIFCONF) { |
1049 | struct ifconf ifc; | |
1050 | if (copy_from_user(&ifc, argp, sizeof(struct ifconf))) | |
1051 | return -EFAULT; | |
1052 | rtnl_lock(); | |
1053 | err = dev_ifconf(net, &ifc, sizeof(struct ifreq)); | |
1054 | rtnl_unlock(); | |
1055 | if (!err && copy_to_user(argp, &ifc, sizeof(struct ifconf))) | |
1056 | err = -EFAULT; | |
44c02a2c AV |
1057 | } else { |
1058 | struct ifreq ifr; | |
1059 | bool need_copyout; | |
63ff03ab | 1060 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) |
44c02a2c AV |
1061 | return -EFAULT; |
1062 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1063 | if (!err && need_copyout) | |
63ff03ab | 1064 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) |
44c02a2c | 1065 | return -EFAULT; |
36fd633e | 1066 | } |
6b96018b AB |
1067 | return err; |
1068 | } | |
1069 | ||
1da177e4 LT |
1070 | /* |
1071 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1072 | * what to do with it - that's up to the protocol still. | |
1073 | */ | |
1074 | ||
8a3c245c PT |
1075 | /** |
1076 | * get_net_ns - increment the refcount of the network namespace | |
1077 | * @ns: common namespace (net) | |
1078 | * | |
1079 | * Returns the net's common namespace. | |
1080 | */ | |
1081 | ||
d8d211a2 | 1082 | struct ns_common *get_net_ns(struct ns_common *ns) |
c62cce2c AV |
1083 | { |
1084 | return &get_net(container_of(ns, struct net, ns))->ns; | |
1085 | } | |
d8d211a2 | 1086 | EXPORT_SYMBOL_GPL(get_net_ns); |
c62cce2c | 1087 | |
1da177e4 LT |
1088 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
1089 | { | |
1090 | struct socket *sock; | |
881d966b | 1091 | struct sock *sk; |
1da177e4 LT |
1092 | void __user *argp = (void __user *)arg; |
1093 | int pid, err; | |
881d966b | 1094 | struct net *net; |
1da177e4 | 1095 | |
b69aee04 | 1096 | sock = file->private_data; |
881d966b | 1097 | sk = sock->sk; |
3b1e0a65 | 1098 | net = sock_net(sk); |
44c02a2c AV |
1099 | if (unlikely(cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))) { |
1100 | struct ifreq ifr; | |
1101 | bool need_copyout; | |
1102 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
1103 | return -EFAULT; | |
1104 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1105 | if (!err && need_copyout) | |
1106 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
1107 | return -EFAULT; | |
1da177e4 | 1108 | } else |
3d23e349 | 1109 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1110 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
b1b0c245 | 1111 | err = wext_handle_ioctl(net, cmd, argp); |
1da177e4 | 1112 | } else |
3d23e349 | 1113 | #endif |
89bddce5 | 1114 | switch (cmd) { |
1da177e4 LT |
1115 | case FIOSETOWN: |
1116 | case SIOCSPGRP: | |
1117 | err = -EFAULT; | |
1118 | if (get_user(pid, (int __user *)argp)) | |
1119 | break; | |
393cc3f5 | 1120 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1121 | break; |
1122 | case FIOGETOWN: | |
1123 | case SIOCGPGRP: | |
609d7fa9 | 1124 | err = put_user(f_getown(sock->file), |
89bddce5 | 1125 | (int __user *)argp); |
1da177e4 LT |
1126 | break; |
1127 | case SIOCGIFBR: | |
1128 | case SIOCSIFBR: | |
1129 | case SIOCBRADDBR: | |
1130 | case SIOCBRDELBR: | |
1131 | err = -ENOPKG; | |
1132 | if (!br_ioctl_hook) | |
1133 | request_module("bridge"); | |
1134 | ||
4a3e2f71 | 1135 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1136 | if (br_ioctl_hook) |
881d966b | 1137 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1138 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1139 | break; |
1140 | case SIOCGIFVLAN: | |
1141 | case SIOCSIFVLAN: | |
1142 | err = -ENOPKG; | |
1143 | if (!vlan_ioctl_hook) | |
1144 | request_module("8021q"); | |
1145 | ||
4a3e2f71 | 1146 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1147 | if (vlan_ioctl_hook) |
881d966b | 1148 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1149 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1150 | break; |
1da177e4 LT |
1151 | case SIOCADDDLCI: |
1152 | case SIOCDELDLCI: | |
1153 | err = -ENOPKG; | |
1154 | if (!dlci_ioctl_hook) | |
1155 | request_module("dlci"); | |
1156 | ||
7512cbf6 PE |
1157 | mutex_lock(&dlci_ioctl_mutex); |
1158 | if (dlci_ioctl_hook) | |
1da177e4 | 1159 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1160 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 | 1161 | break; |
c62cce2c AV |
1162 | case SIOCGSKNS: |
1163 | err = -EPERM; | |
1164 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1165 | break; | |
1166 | ||
1167 | err = open_related_ns(&net->ns, get_net_ns); | |
1168 | break; | |
0768e170 AB |
1169 | case SIOCGSTAMP_OLD: |
1170 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
1171 | if (!sock->ops->gettstamp) { |
1172 | err = -ENOIOCTLCMD; | |
1173 | break; | |
1174 | } | |
1175 | err = sock->ops->gettstamp(sock, argp, | |
0768e170 AB |
1176 | cmd == SIOCGSTAMP_OLD, |
1177 | !IS_ENABLED(CONFIG_64BIT)); | |
60747828 | 1178 | break; |
0768e170 AB |
1179 | case SIOCGSTAMP_NEW: |
1180 | case SIOCGSTAMPNS_NEW: | |
1181 | if (!sock->ops->gettstamp) { | |
1182 | err = -ENOIOCTLCMD; | |
1183 | break; | |
1184 | } | |
1185 | err = sock->ops->gettstamp(sock, argp, | |
1186 | cmd == SIOCGSTAMP_NEW, | |
1187 | false); | |
c7cbdbf2 | 1188 | break; |
1da177e4 | 1189 | default: |
63ff03ab | 1190 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1191 | break; |
89bddce5 | 1192 | } |
1da177e4 LT |
1193 | return err; |
1194 | } | |
1195 | ||
8a3c245c PT |
1196 | /** |
1197 | * sock_create_lite - creates a socket | |
1198 | * @family: protocol family (AF_INET, ...) | |
1199 | * @type: communication type (SOCK_STREAM, ...) | |
1200 | * @protocol: protocol (0, ...) | |
1201 | * @res: new socket | |
1202 | * | |
1203 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1204 | * The new socket initialization is not complete, see kernel_accept(). | |
1205 | * Returns 0 or an error. On failure @res is set to %NULL. | |
1206 | * This function internally uses GFP_KERNEL. | |
1207 | */ | |
1208 | ||
1da177e4 LT |
1209 | int sock_create_lite(int family, int type, int protocol, struct socket **res) |
1210 | { | |
1211 | int err; | |
1212 | struct socket *sock = NULL; | |
89bddce5 | 1213 | |
1da177e4 LT |
1214 | err = security_socket_create(family, type, protocol, 1); |
1215 | if (err) | |
1216 | goto out; | |
1217 | ||
1218 | sock = sock_alloc(); | |
1219 | if (!sock) { | |
1220 | err = -ENOMEM; | |
1221 | goto out; | |
1222 | } | |
1223 | ||
1da177e4 | 1224 | sock->type = type; |
7420ed23 VY |
1225 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1226 | if (err) | |
1227 | goto out_release; | |
1228 | ||
1da177e4 LT |
1229 | out: |
1230 | *res = sock; | |
1231 | return err; | |
7420ed23 VY |
1232 | out_release: |
1233 | sock_release(sock); | |
1234 | sock = NULL; | |
1235 | goto out; | |
1da177e4 | 1236 | } |
c6d409cf | 1237 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1238 | |
1239 | /* No kernel lock held - perfect */ | |
ade994f4 | 1240 | static __poll_t sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1241 | { |
3cafb376 | 1242 | struct socket *sock = file->private_data; |
a331de3b | 1243 | __poll_t events = poll_requested_events(wait), flag = 0; |
2d48d67f | 1244 | |
e88958e6 CH |
1245 | if (!sock->ops->poll) |
1246 | return 0; | |
f641f13b | 1247 | |
a331de3b CH |
1248 | if (sk_can_busy_loop(sock->sk)) { |
1249 | /* poll once if requested by the syscall */ | |
1250 | if (events & POLL_BUSY_LOOP) | |
1251 | sk_busy_loop(sock->sk, 1); | |
1252 | ||
1253 | /* if this socket can poll_ll, tell the system call */ | |
1254 | flag = POLL_BUSY_LOOP; | |
1255 | } | |
1256 | ||
1257 | return sock->ops->poll(file, sock, wait) | flag; | |
1da177e4 LT |
1258 | } |
1259 | ||
89bddce5 | 1260 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1261 | { |
b69aee04 | 1262 | struct socket *sock = file->private_data; |
1da177e4 LT |
1263 | |
1264 | return sock->ops->mmap(file, sock, vma); | |
1265 | } | |
1266 | ||
20380731 | 1267 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1268 | { |
6d8c50dc | 1269 | __sock_release(SOCKET_I(inode), inode); |
1da177e4 LT |
1270 | return 0; |
1271 | } | |
1272 | ||
1273 | /* | |
1274 | * Update the socket async list | |
1275 | * | |
1276 | * Fasync_list locking strategy. | |
1277 | * | |
1278 | * 1. fasync_list is modified only under process context socket lock | |
1279 | * i.e. under semaphore. | |
1280 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1281 | * or under socket lock |
1da177e4 LT |
1282 | */ |
1283 | ||
1284 | static int sock_fasync(int fd, struct file *filp, int on) | |
1285 | { | |
989a2979 ED |
1286 | struct socket *sock = filp->private_data; |
1287 | struct sock *sk = sock->sk; | |
333f7909 | 1288 | struct socket_wq *wq = &sock->wq; |
1da177e4 | 1289 | |
989a2979 | 1290 | if (sk == NULL) |
1da177e4 | 1291 | return -EINVAL; |
1da177e4 LT |
1292 | |
1293 | lock_sock(sk); | |
eaefd110 | 1294 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1295 | |
eaefd110 | 1296 | if (!wq->fasync_list) |
989a2979 ED |
1297 | sock_reset_flag(sk, SOCK_FASYNC); |
1298 | else | |
bcdce719 | 1299 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1300 | |
989a2979 | 1301 | release_sock(sk); |
1da177e4 LT |
1302 | return 0; |
1303 | } | |
1304 | ||
ceb5d58b | 1305 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1306 | |
ceb5d58b | 1307 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1308 | { |
ceb5d58b | 1309 | if (!wq || !wq->fasync_list) |
1da177e4 | 1310 | return -1; |
ceb5d58b | 1311 | |
89bddce5 | 1312 | switch (how) { |
8d8ad9d7 | 1313 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1314 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1315 | break; |
1316 | goto call_kill; | |
8d8ad9d7 | 1317 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1318 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 LT |
1319 | break; |
1320 | /* fall through */ | |
8d8ad9d7 | 1321 | case SOCK_WAKE_IO: |
89bddce5 | 1322 | call_kill: |
43815482 | 1323 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1324 | break; |
8d8ad9d7 | 1325 | case SOCK_WAKE_URG: |
43815482 | 1326 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1327 | } |
ceb5d58b | 1328 | |
1da177e4 LT |
1329 | return 0; |
1330 | } | |
c6d409cf | 1331 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1332 | |
8a3c245c PT |
1333 | /** |
1334 | * __sock_create - creates a socket | |
1335 | * @net: net namespace | |
1336 | * @family: protocol family (AF_INET, ...) | |
1337 | * @type: communication type (SOCK_STREAM, ...) | |
1338 | * @protocol: protocol (0, ...) | |
1339 | * @res: new socket | |
1340 | * @kern: boolean for kernel space sockets | |
1341 | * | |
1342 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1343 | * Returns 0 or an error. On failure @res is set to %NULL. @kern must | |
1344 | * be set to true if the socket resides in kernel space. | |
1345 | * This function internally uses GFP_KERNEL. | |
1346 | */ | |
1347 | ||
721db93a | 1348 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1349 | struct socket **res, int kern) |
1da177e4 LT |
1350 | { |
1351 | int err; | |
1352 | struct socket *sock; | |
55737fda | 1353 | const struct net_proto_family *pf; |
1da177e4 LT |
1354 | |
1355 | /* | |
89bddce5 | 1356 | * Check protocol is in range |
1da177e4 LT |
1357 | */ |
1358 | if (family < 0 || family >= NPROTO) | |
1359 | return -EAFNOSUPPORT; | |
1360 | if (type < 0 || type >= SOCK_MAX) | |
1361 | return -EINVAL; | |
1362 | ||
1363 | /* Compatibility. | |
1364 | ||
1365 | This uglymoron is moved from INET layer to here to avoid | |
1366 | deadlock in module load. | |
1367 | */ | |
1368 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1369 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1370 | current->comm); | |
1da177e4 LT |
1371 | family = PF_PACKET; |
1372 | } | |
1373 | ||
1374 | err = security_socket_create(family, type, protocol, kern); | |
1375 | if (err) | |
1376 | return err; | |
89bddce5 | 1377 | |
55737fda SH |
1378 | /* |
1379 | * Allocate the socket and allow the family to set things up. if | |
1380 | * the protocol is 0, the family is instructed to select an appropriate | |
1381 | * default. | |
1382 | */ | |
1383 | sock = sock_alloc(); | |
1384 | if (!sock) { | |
e87cc472 | 1385 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1386 | return -ENFILE; /* Not exactly a match, but its the |
1387 | closest posix thing */ | |
1388 | } | |
1389 | ||
1390 | sock->type = type; | |
1391 | ||
95a5afca | 1392 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1393 | /* Attempt to load a protocol module if the find failed. |
1394 | * | |
1395 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1396 | * requested real, full-featured networking support upon configuration. |
1397 | * Otherwise module support will break! | |
1398 | */ | |
190683a9 | 1399 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1400 | request_module("net-pf-%d", family); |
1da177e4 LT |
1401 | #endif |
1402 | ||
55737fda SH |
1403 | rcu_read_lock(); |
1404 | pf = rcu_dereference(net_families[family]); | |
1405 | err = -EAFNOSUPPORT; | |
1406 | if (!pf) | |
1407 | goto out_release; | |
1da177e4 LT |
1408 | |
1409 | /* | |
1410 | * We will call the ->create function, that possibly is in a loadable | |
1411 | * module, so we have to bump that loadable module refcnt first. | |
1412 | */ | |
55737fda | 1413 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1414 | goto out_release; |
1415 | ||
55737fda SH |
1416 | /* Now protected by module ref count */ |
1417 | rcu_read_unlock(); | |
1418 | ||
3f378b68 | 1419 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1420 | if (err < 0) |
1da177e4 | 1421 | goto out_module_put; |
a79af59e | 1422 | |
1da177e4 LT |
1423 | /* |
1424 | * Now to bump the refcnt of the [loadable] module that owns this | |
1425 | * socket at sock_release time we decrement its refcnt. | |
1426 | */ | |
55737fda SH |
1427 | if (!try_module_get(sock->ops->owner)) |
1428 | goto out_module_busy; | |
1429 | ||
1da177e4 LT |
1430 | /* |
1431 | * Now that we're done with the ->create function, the [loadable] | |
1432 | * module can have its refcnt decremented | |
1433 | */ | |
55737fda | 1434 | module_put(pf->owner); |
7420ed23 VY |
1435 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1436 | if (err) | |
3b185525 | 1437 | goto out_sock_release; |
55737fda | 1438 | *res = sock; |
1da177e4 | 1439 | |
55737fda SH |
1440 | return 0; |
1441 | ||
1442 | out_module_busy: | |
1443 | err = -EAFNOSUPPORT; | |
1da177e4 | 1444 | out_module_put: |
55737fda SH |
1445 | sock->ops = NULL; |
1446 | module_put(pf->owner); | |
1447 | out_sock_release: | |
1da177e4 | 1448 | sock_release(sock); |
55737fda SH |
1449 | return err; |
1450 | ||
1451 | out_release: | |
1452 | rcu_read_unlock(); | |
1453 | goto out_sock_release; | |
1da177e4 | 1454 | } |
721db93a | 1455 | EXPORT_SYMBOL(__sock_create); |
1da177e4 | 1456 | |
8a3c245c PT |
1457 | /** |
1458 | * sock_create - creates a socket | |
1459 | * @family: protocol family (AF_INET, ...) | |
1460 | * @type: communication type (SOCK_STREAM, ...) | |
1461 | * @protocol: protocol (0, ...) | |
1462 | * @res: new socket | |
1463 | * | |
1464 | * A wrapper around __sock_create(). | |
1465 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1466 | */ | |
1467 | ||
1da177e4 LT |
1468 | int sock_create(int family, int type, int protocol, struct socket **res) |
1469 | { | |
1b8d7ae4 | 1470 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1471 | } |
c6d409cf | 1472 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1473 | |
8a3c245c PT |
1474 | /** |
1475 | * sock_create_kern - creates a socket (kernel space) | |
1476 | * @net: net namespace | |
1477 | * @family: protocol family (AF_INET, ...) | |
1478 | * @type: communication type (SOCK_STREAM, ...) | |
1479 | * @protocol: protocol (0, ...) | |
1480 | * @res: new socket | |
1481 | * | |
1482 | * A wrapper around __sock_create(). | |
1483 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1484 | */ | |
1485 | ||
eeb1bd5c | 1486 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1487 | { |
eeb1bd5c | 1488 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1489 | } |
c6d409cf | 1490 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1491 | |
9d6a15c3 | 1492 | int __sys_socket(int family, int type, int protocol) |
1da177e4 LT |
1493 | { |
1494 | int retval; | |
1495 | struct socket *sock; | |
a677a039 UD |
1496 | int flags; |
1497 | ||
e38b36f3 UD |
1498 | /* Check the SOCK_* constants for consistency. */ |
1499 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1500 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1501 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1502 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1503 | ||
a677a039 | 1504 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1505 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1506 | return -EINVAL; |
1507 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1508 | |
aaca0bdc UD |
1509 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1510 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1511 | ||
1da177e4 LT |
1512 | retval = sock_create(family, type, protocol, &sock); |
1513 | if (retval < 0) | |
8e1611e2 | 1514 | return retval; |
1da177e4 | 1515 | |
8e1611e2 | 1516 | return sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1517 | } |
1518 | ||
9d6a15c3 DB |
1519 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1520 | { | |
1521 | return __sys_socket(family, type, protocol); | |
1522 | } | |
1523 | ||
1da177e4 LT |
1524 | /* |
1525 | * Create a pair of connected sockets. | |
1526 | */ | |
1527 | ||
6debc8d8 | 1528 | int __sys_socketpair(int family, int type, int protocol, int __user *usockvec) |
1da177e4 LT |
1529 | { |
1530 | struct socket *sock1, *sock2; | |
1531 | int fd1, fd2, err; | |
db349509 | 1532 | struct file *newfile1, *newfile2; |
a677a039 UD |
1533 | int flags; |
1534 | ||
1535 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1536 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1537 | return -EINVAL; |
1538 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1539 | |
aaca0bdc UD |
1540 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1541 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1542 | ||
016a266b AV |
1543 | /* |
1544 | * reserve descriptors and make sure we won't fail | |
1545 | * to return them to userland. | |
1546 | */ | |
1547 | fd1 = get_unused_fd_flags(flags); | |
1548 | if (unlikely(fd1 < 0)) | |
1549 | return fd1; | |
1550 | ||
1551 | fd2 = get_unused_fd_flags(flags); | |
1552 | if (unlikely(fd2 < 0)) { | |
1553 | put_unused_fd(fd1); | |
1554 | return fd2; | |
1555 | } | |
1556 | ||
1557 | err = put_user(fd1, &usockvec[0]); | |
1558 | if (err) | |
1559 | goto out; | |
1560 | ||
1561 | err = put_user(fd2, &usockvec[1]); | |
1562 | if (err) | |
1563 | goto out; | |
1564 | ||
1da177e4 LT |
1565 | /* |
1566 | * Obtain the first socket and check if the underlying protocol | |
1567 | * supports the socketpair call. | |
1568 | */ | |
1569 | ||
1570 | err = sock_create(family, type, protocol, &sock1); | |
016a266b | 1571 | if (unlikely(err < 0)) |
1da177e4 LT |
1572 | goto out; |
1573 | ||
1574 | err = sock_create(family, type, protocol, &sock2); | |
016a266b AV |
1575 | if (unlikely(err < 0)) { |
1576 | sock_release(sock1); | |
1577 | goto out; | |
bf3c23d1 | 1578 | } |
d73aa286 | 1579 | |
d47cd945 DH |
1580 | err = security_socket_socketpair(sock1, sock2); |
1581 | if (unlikely(err)) { | |
1582 | sock_release(sock2); | |
1583 | sock_release(sock1); | |
1584 | goto out; | |
1585 | } | |
1586 | ||
016a266b AV |
1587 | err = sock1->ops->socketpair(sock1, sock2); |
1588 | if (unlikely(err < 0)) { | |
1589 | sock_release(sock2); | |
1590 | sock_release(sock1); | |
1591 | goto out; | |
28407630 AV |
1592 | } |
1593 | ||
aab174f0 | 1594 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1595 | if (IS_ERR(newfile1)) { |
28407630 | 1596 | err = PTR_ERR(newfile1); |
016a266b AV |
1597 | sock_release(sock2); |
1598 | goto out; | |
28407630 AV |
1599 | } |
1600 | ||
aab174f0 | 1601 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1602 | if (IS_ERR(newfile2)) { |
1603 | err = PTR_ERR(newfile2); | |
016a266b AV |
1604 | fput(newfile1); |
1605 | goto out; | |
db349509 AV |
1606 | } |
1607 | ||
157cf649 | 1608 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1609 | |
db349509 AV |
1610 | fd_install(fd1, newfile1); |
1611 | fd_install(fd2, newfile2); | |
d73aa286 | 1612 | return 0; |
1da177e4 | 1613 | |
016a266b | 1614 | out: |
d73aa286 | 1615 | put_unused_fd(fd2); |
d73aa286 | 1616 | put_unused_fd(fd1); |
1da177e4 LT |
1617 | return err; |
1618 | } | |
1619 | ||
6debc8d8 DB |
1620 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1621 | int __user *, usockvec) | |
1622 | { | |
1623 | return __sys_socketpair(family, type, protocol, usockvec); | |
1624 | } | |
1625 | ||
1da177e4 LT |
1626 | /* |
1627 | * Bind a name to a socket. Nothing much to do here since it's | |
1628 | * the protocol's responsibility to handle the local address. | |
1629 | * | |
1630 | * We move the socket address to kernel space before we call | |
1631 | * the protocol layer (having also checked the address is ok). | |
1632 | */ | |
1633 | ||
a87d35d8 | 1634 | int __sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
1da177e4 LT |
1635 | { |
1636 | struct socket *sock; | |
230b1839 | 1637 | struct sockaddr_storage address; |
6cb153ca | 1638 | int err, fput_needed; |
1da177e4 | 1639 | |
89bddce5 | 1640 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1641 | if (sock) { |
43db362d | 1642 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
068b88cc | 1643 | if (!err) { |
89bddce5 | 1644 | err = security_socket_bind(sock, |
230b1839 | 1645 | (struct sockaddr *)&address, |
89bddce5 | 1646 | addrlen); |
6cb153ca BL |
1647 | if (!err) |
1648 | err = sock->ops->bind(sock, | |
89bddce5 | 1649 | (struct sockaddr *) |
230b1839 | 1650 | &address, addrlen); |
1da177e4 | 1651 | } |
6cb153ca | 1652 | fput_light(sock->file, fput_needed); |
89bddce5 | 1653 | } |
1da177e4 LT |
1654 | return err; |
1655 | } | |
1656 | ||
a87d35d8 DB |
1657 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1658 | { | |
1659 | return __sys_bind(fd, umyaddr, addrlen); | |
1660 | } | |
1661 | ||
1da177e4 LT |
1662 | /* |
1663 | * Perform a listen. Basically, we allow the protocol to do anything | |
1664 | * necessary for a listen, and if that works, we mark the socket as | |
1665 | * ready for listening. | |
1666 | */ | |
1667 | ||
25e290ee | 1668 | int __sys_listen(int fd, int backlog) |
1da177e4 LT |
1669 | { |
1670 | struct socket *sock; | |
6cb153ca | 1671 | int err, fput_needed; |
b8e1f9b5 | 1672 | int somaxconn; |
89bddce5 SH |
1673 | |
1674 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1675 | if (sock) { | |
8efa6e93 | 1676 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1677 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1678 | backlog = somaxconn; |
1da177e4 LT |
1679 | |
1680 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1681 | if (!err) |
1682 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1683 | |
6cb153ca | 1684 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1685 | } |
1686 | return err; | |
1687 | } | |
1688 | ||
25e290ee DB |
1689 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1690 | { | |
1691 | return __sys_listen(fd, backlog); | |
1692 | } | |
1693 | ||
de2ea4b6 JA |
1694 | int __sys_accept4_file(struct file *file, unsigned file_flags, |
1695 | struct sockaddr __user *upeer_sockaddr, | |
1696 | int __user *upeer_addrlen, int flags) | |
1da177e4 LT |
1697 | { |
1698 | struct socket *sock, *newsock; | |
39d8c1b6 | 1699 | struct file *newfile; |
de2ea4b6 | 1700 | int err, len, newfd; |
230b1839 | 1701 | struct sockaddr_storage address; |
1da177e4 | 1702 | |
77d27200 | 1703 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1704 | return -EINVAL; |
1705 | ||
1706 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1707 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1708 | ||
de2ea4b6 | 1709 | sock = sock_from_file(file, &err); |
1da177e4 LT |
1710 | if (!sock) |
1711 | goto out; | |
1712 | ||
1713 | err = -ENFILE; | |
c6d409cf ED |
1714 | newsock = sock_alloc(); |
1715 | if (!newsock) | |
de2ea4b6 | 1716 | goto out; |
1da177e4 LT |
1717 | |
1718 | newsock->type = sock->type; | |
1719 | newsock->ops = sock->ops; | |
1720 | ||
1da177e4 LT |
1721 | /* |
1722 | * We don't need try_module_get here, as the listening socket (sock) | |
1723 | * has the protocol module (sock->ops->owner) held. | |
1724 | */ | |
1725 | __module_get(newsock->ops->owner); | |
1726 | ||
28407630 | 1727 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1728 | if (unlikely(newfd < 0)) { |
1729 | err = newfd; | |
9a1875e6 | 1730 | sock_release(newsock); |
de2ea4b6 | 1731 | goto out; |
39d8c1b6 | 1732 | } |
aab174f0 | 1733 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1734 | if (IS_ERR(newfile)) { |
28407630 AV |
1735 | err = PTR_ERR(newfile); |
1736 | put_unused_fd(newfd); | |
de2ea4b6 | 1737 | goto out; |
28407630 | 1738 | } |
39d8c1b6 | 1739 | |
a79af59e FF |
1740 | err = security_socket_accept(sock, newsock); |
1741 | if (err) | |
39d8c1b6 | 1742 | goto out_fd; |
a79af59e | 1743 | |
de2ea4b6 JA |
1744 | err = sock->ops->accept(sock, newsock, sock->file->f_flags | file_flags, |
1745 | false); | |
1da177e4 | 1746 | if (err < 0) |
39d8c1b6 | 1747 | goto out_fd; |
1da177e4 LT |
1748 | |
1749 | if (upeer_sockaddr) { | |
9b2c45d4 DV |
1750 | len = newsock->ops->getname(newsock, |
1751 | (struct sockaddr *)&address, 2); | |
1752 | if (len < 0) { | |
1da177e4 | 1753 | err = -ECONNABORTED; |
39d8c1b6 | 1754 | goto out_fd; |
1da177e4 | 1755 | } |
43db362d | 1756 | err = move_addr_to_user(&address, |
230b1839 | 1757 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1758 | if (err < 0) |
39d8c1b6 | 1759 | goto out_fd; |
1da177e4 LT |
1760 | } |
1761 | ||
1762 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1763 | ||
39d8c1b6 DM |
1764 | fd_install(newfd, newfile); |
1765 | err = newfd; | |
1da177e4 LT |
1766 | out: |
1767 | return err; | |
39d8c1b6 | 1768 | out_fd: |
9606a216 | 1769 | fput(newfile); |
39d8c1b6 | 1770 | put_unused_fd(newfd); |
de2ea4b6 JA |
1771 | goto out; |
1772 | ||
1773 | } | |
1774 | ||
1775 | /* | |
1776 | * For accept, we attempt to create a new socket, set up the link | |
1777 | * with the client, wake up the client, then return the new | |
1778 | * connected fd. We collect the address of the connector in kernel | |
1779 | * space and move it to user at the very end. This is unclean because | |
1780 | * we open the socket then return an error. | |
1781 | * | |
1782 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1783 | * status to recvmsg. We need to add that support in a way thats | |
1784 | * clean when we restructure accept also. | |
1785 | */ | |
1786 | ||
1787 | int __sys_accept4(int fd, struct sockaddr __user *upeer_sockaddr, | |
1788 | int __user *upeer_addrlen, int flags) | |
1789 | { | |
1790 | int ret = -EBADF; | |
1791 | struct fd f; | |
1792 | ||
1793 | f = fdget(fd); | |
1794 | if (f.file) { | |
1795 | ret = __sys_accept4_file(f.file, 0, upeer_sockaddr, | |
1796 | upeer_addrlen, flags); | |
1797 | if (f.flags) | |
1798 | fput(f.file); | |
1799 | } | |
1800 | ||
1801 | return ret; | |
1da177e4 LT |
1802 | } |
1803 | ||
4541e805 DB |
1804 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1805 | int __user *, upeer_addrlen, int, flags) | |
1806 | { | |
1807 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, flags); | |
1808 | } | |
1809 | ||
20f37034 HC |
1810 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1811 | int __user *, upeer_addrlen) | |
aaca0bdc | 1812 | { |
4541e805 | 1813 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1814 | } |
1815 | ||
1da177e4 LT |
1816 | /* |
1817 | * Attempt to connect to a socket with the server address. The address | |
1818 | * is in user space so we verify it is OK and move it to kernel space. | |
1819 | * | |
1820 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1821 | * break bindings | |
1822 | * | |
1823 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1824 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1825 | * include the -EINPROGRESS status for such sockets. | |
1826 | */ | |
1827 | ||
1387c2c2 | 1828 | int __sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
1da177e4 LT |
1829 | { |
1830 | struct socket *sock; | |
230b1839 | 1831 | struct sockaddr_storage address; |
6cb153ca | 1832 | int err, fput_needed; |
1da177e4 | 1833 | |
6cb153ca | 1834 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1835 | if (!sock) |
1836 | goto out; | |
43db362d | 1837 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1838 | if (err < 0) |
1839 | goto out_put; | |
1840 | ||
89bddce5 | 1841 | err = |
230b1839 | 1842 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1843 | if (err) |
1844 | goto out_put; | |
1845 | ||
230b1839 | 1846 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1847 | sock->file->f_flags); |
1848 | out_put: | |
6cb153ca | 1849 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1850 | out: |
1851 | return err; | |
1852 | } | |
1853 | ||
1387c2c2 DB |
1854 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1855 | int, addrlen) | |
1856 | { | |
1857 | return __sys_connect(fd, uservaddr, addrlen); | |
1858 | } | |
1859 | ||
1da177e4 LT |
1860 | /* |
1861 | * Get the local address ('name') of a socket object. Move the obtained | |
1862 | * name to user space. | |
1863 | */ | |
1864 | ||
8882a107 DB |
1865 | int __sys_getsockname(int fd, struct sockaddr __user *usockaddr, |
1866 | int __user *usockaddr_len) | |
1da177e4 LT |
1867 | { |
1868 | struct socket *sock; | |
230b1839 | 1869 | struct sockaddr_storage address; |
9b2c45d4 | 1870 | int err, fput_needed; |
89bddce5 | 1871 | |
6cb153ca | 1872 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1873 | if (!sock) |
1874 | goto out; | |
1875 | ||
1876 | err = security_socket_getsockname(sock); | |
1877 | if (err) | |
1878 | goto out_put; | |
1879 | ||
9b2c45d4 DV |
1880 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 0); |
1881 | if (err < 0) | |
1da177e4 | 1882 | goto out_put; |
9b2c45d4 DV |
1883 | /* "err" is actually length in this case */ |
1884 | err = move_addr_to_user(&address, err, usockaddr, usockaddr_len); | |
1da177e4 LT |
1885 | |
1886 | out_put: | |
6cb153ca | 1887 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1888 | out: |
1889 | return err; | |
1890 | } | |
1891 | ||
8882a107 DB |
1892 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1893 | int __user *, usockaddr_len) | |
1894 | { | |
1895 | return __sys_getsockname(fd, usockaddr, usockaddr_len); | |
1896 | } | |
1897 | ||
1da177e4 LT |
1898 | /* |
1899 | * Get the remote address ('name') of a socket object. Move the obtained | |
1900 | * name to user space. | |
1901 | */ | |
1902 | ||
b21c8f83 DB |
1903 | int __sys_getpeername(int fd, struct sockaddr __user *usockaddr, |
1904 | int __user *usockaddr_len) | |
1da177e4 LT |
1905 | { |
1906 | struct socket *sock; | |
230b1839 | 1907 | struct sockaddr_storage address; |
9b2c45d4 | 1908 | int err, fput_needed; |
1da177e4 | 1909 | |
89bddce5 SH |
1910 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1911 | if (sock != NULL) { | |
1da177e4 LT |
1912 | err = security_socket_getpeername(sock); |
1913 | if (err) { | |
6cb153ca | 1914 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1915 | return err; |
1916 | } | |
1917 | ||
9b2c45d4 DV |
1918 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 1); |
1919 | if (err >= 0) | |
1920 | /* "err" is actually length in this case */ | |
1921 | err = move_addr_to_user(&address, err, usockaddr, | |
89bddce5 | 1922 | usockaddr_len); |
6cb153ca | 1923 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1924 | } |
1925 | return err; | |
1926 | } | |
1927 | ||
b21c8f83 DB |
1928 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1929 | int __user *, usockaddr_len) | |
1930 | { | |
1931 | return __sys_getpeername(fd, usockaddr, usockaddr_len); | |
1932 | } | |
1933 | ||
1da177e4 LT |
1934 | /* |
1935 | * Send a datagram to a given address. We move the address into kernel | |
1936 | * space and check the user space data area is readable before invoking | |
1937 | * the protocol. | |
1938 | */ | |
211b634b DB |
1939 | int __sys_sendto(int fd, void __user *buff, size_t len, unsigned int flags, |
1940 | struct sockaddr __user *addr, int addr_len) | |
1da177e4 LT |
1941 | { |
1942 | struct socket *sock; | |
230b1839 | 1943 | struct sockaddr_storage address; |
1da177e4 LT |
1944 | int err; |
1945 | struct msghdr msg; | |
1946 | struct iovec iov; | |
6cb153ca | 1947 | int fput_needed; |
6cb153ca | 1948 | |
602bd0e9 AV |
1949 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1950 | if (unlikely(err)) | |
1951 | return err; | |
de0fa95c PE |
1952 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1953 | if (!sock) | |
4387ff75 | 1954 | goto out; |
6cb153ca | 1955 | |
89bddce5 | 1956 | msg.msg_name = NULL; |
89bddce5 SH |
1957 | msg.msg_control = NULL; |
1958 | msg.msg_controllen = 0; | |
1959 | msg.msg_namelen = 0; | |
6cb153ca | 1960 | if (addr) { |
43db362d | 1961 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1962 | if (err < 0) |
1963 | goto out_put; | |
230b1839 | 1964 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1965 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1966 | } |
1967 | if (sock->file->f_flags & O_NONBLOCK) | |
1968 | flags |= MSG_DONTWAIT; | |
1969 | msg.msg_flags = flags; | |
d8725c86 | 1970 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 1971 | |
89bddce5 | 1972 | out_put: |
de0fa95c | 1973 | fput_light(sock->file, fput_needed); |
4387ff75 | 1974 | out: |
1da177e4 LT |
1975 | return err; |
1976 | } | |
1977 | ||
211b634b DB |
1978 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
1979 | unsigned int, flags, struct sockaddr __user *, addr, | |
1980 | int, addr_len) | |
1981 | { | |
1982 | return __sys_sendto(fd, buff, len, flags, addr, addr_len); | |
1983 | } | |
1984 | ||
1da177e4 | 1985 | /* |
89bddce5 | 1986 | * Send a datagram down a socket. |
1da177e4 LT |
1987 | */ |
1988 | ||
3e0fa65f | 1989 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1990 | unsigned int, flags) |
1da177e4 | 1991 | { |
211b634b | 1992 | return __sys_sendto(fd, buff, len, flags, NULL, 0); |
1da177e4 LT |
1993 | } |
1994 | ||
1995 | /* | |
89bddce5 | 1996 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1997 | * sender. We verify the buffers are writable and if needed move the |
1998 | * sender address from kernel to user space. | |
1999 | */ | |
7a09e1eb DB |
2000 | int __sys_recvfrom(int fd, void __user *ubuf, size_t size, unsigned int flags, |
2001 | struct sockaddr __user *addr, int __user *addr_len) | |
1da177e4 LT |
2002 | { |
2003 | struct socket *sock; | |
2004 | struct iovec iov; | |
2005 | struct msghdr msg; | |
230b1839 | 2006 | struct sockaddr_storage address; |
89bddce5 | 2007 | int err, err2; |
6cb153ca BL |
2008 | int fput_needed; |
2009 | ||
602bd0e9 AV |
2010 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
2011 | if (unlikely(err)) | |
2012 | return err; | |
de0fa95c | 2013 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 2014 | if (!sock) |
de0fa95c | 2015 | goto out; |
1da177e4 | 2016 | |
89bddce5 SH |
2017 | msg.msg_control = NULL; |
2018 | msg.msg_controllen = 0; | |
f3d33426 HFS |
2019 | /* Save some cycles and don't copy the address if not needed */ |
2020 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
2021 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
2022 | msg.msg_namelen = 0; | |
130ed5d1 | 2023 | msg.msg_iocb = NULL; |
9f138fa6 | 2024 | msg.msg_flags = 0; |
1da177e4 LT |
2025 | if (sock->file->f_flags & O_NONBLOCK) |
2026 | flags |= MSG_DONTWAIT; | |
2da62906 | 2027 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 2028 | |
89bddce5 | 2029 | if (err >= 0 && addr != NULL) { |
43db362d | 2030 | err2 = move_addr_to_user(&address, |
230b1839 | 2031 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
2032 | if (err2 < 0) |
2033 | err = err2; | |
1da177e4 | 2034 | } |
de0fa95c PE |
2035 | |
2036 | fput_light(sock->file, fput_needed); | |
4387ff75 | 2037 | out: |
1da177e4 LT |
2038 | return err; |
2039 | } | |
2040 | ||
7a09e1eb DB |
2041 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
2042 | unsigned int, flags, struct sockaddr __user *, addr, | |
2043 | int __user *, addr_len) | |
2044 | { | |
2045 | return __sys_recvfrom(fd, ubuf, size, flags, addr, addr_len); | |
2046 | } | |
2047 | ||
1da177e4 | 2048 | /* |
89bddce5 | 2049 | * Receive a datagram from a socket. |
1da177e4 LT |
2050 | */ |
2051 | ||
b7c0ddf5 JG |
2052 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
2053 | unsigned int, flags) | |
1da177e4 | 2054 | { |
7a09e1eb | 2055 | return __sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
1da177e4 LT |
2056 | } |
2057 | ||
2058 | /* | |
2059 | * Set a socket option. Because we don't know the option lengths we have | |
2060 | * to pass the user mode parameter for the protocols to sort out. | |
2061 | */ | |
2062 | ||
cc36dca0 DB |
2063 | static int __sys_setsockopt(int fd, int level, int optname, |
2064 | char __user *optval, int optlen) | |
1da177e4 | 2065 | { |
0d01da6a SF |
2066 | mm_segment_t oldfs = get_fs(); |
2067 | char *kernel_optval = NULL; | |
6cb153ca | 2068 | int err, fput_needed; |
1da177e4 LT |
2069 | struct socket *sock; |
2070 | ||
2071 | if (optlen < 0) | |
2072 | return -EINVAL; | |
89bddce5 SH |
2073 | |
2074 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2075 | if (sock != NULL) { | |
2076 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
2077 | if (err) |
2078 | goto out_put; | |
1da177e4 | 2079 | |
0d01da6a SF |
2080 | err = BPF_CGROUP_RUN_PROG_SETSOCKOPT(sock->sk, &level, |
2081 | &optname, optval, &optlen, | |
2082 | &kernel_optval); | |
2083 | ||
2084 | if (err < 0) { | |
2085 | goto out_put; | |
2086 | } else if (err > 0) { | |
2087 | err = 0; | |
2088 | goto out_put; | |
2089 | } | |
2090 | ||
2091 | if (kernel_optval) { | |
2092 | set_fs(KERNEL_DS); | |
2093 | optval = (char __user __force *)kernel_optval; | |
2094 | } | |
2095 | ||
1da177e4 | 2096 | if (level == SOL_SOCKET) |
89bddce5 SH |
2097 | err = |
2098 | sock_setsockopt(sock, level, optname, optval, | |
2099 | optlen); | |
1da177e4 | 2100 | else |
89bddce5 SH |
2101 | err = |
2102 | sock->ops->setsockopt(sock, level, optname, optval, | |
2103 | optlen); | |
0d01da6a SF |
2104 | |
2105 | if (kernel_optval) { | |
2106 | set_fs(oldfs); | |
2107 | kfree(kernel_optval); | |
2108 | } | |
6cb153ca BL |
2109 | out_put: |
2110 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2111 | } |
2112 | return err; | |
2113 | } | |
2114 | ||
cc36dca0 DB |
2115 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
2116 | char __user *, optval, int, optlen) | |
2117 | { | |
2118 | return __sys_setsockopt(fd, level, optname, optval, optlen); | |
2119 | } | |
2120 | ||
1da177e4 LT |
2121 | /* |
2122 | * Get a socket option. Because we don't know the option lengths we have | |
2123 | * to pass a user mode parameter for the protocols to sort out. | |
2124 | */ | |
2125 | ||
13a2d70e DB |
2126 | static int __sys_getsockopt(int fd, int level, int optname, |
2127 | char __user *optval, int __user *optlen) | |
1da177e4 | 2128 | { |
6cb153ca | 2129 | int err, fput_needed; |
1da177e4 | 2130 | struct socket *sock; |
0d01da6a | 2131 | int max_optlen; |
1da177e4 | 2132 | |
89bddce5 SH |
2133 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2134 | if (sock != NULL) { | |
6cb153ca BL |
2135 | err = security_socket_getsockopt(sock, level, optname); |
2136 | if (err) | |
2137 | goto out_put; | |
1da177e4 | 2138 | |
0d01da6a SF |
2139 | max_optlen = BPF_CGROUP_GETSOCKOPT_MAX_OPTLEN(optlen); |
2140 | ||
1da177e4 | 2141 | if (level == SOL_SOCKET) |
89bddce5 SH |
2142 | err = |
2143 | sock_getsockopt(sock, level, optname, optval, | |
2144 | optlen); | |
1da177e4 | 2145 | else |
89bddce5 SH |
2146 | err = |
2147 | sock->ops->getsockopt(sock, level, optname, optval, | |
2148 | optlen); | |
0d01da6a SF |
2149 | |
2150 | err = BPF_CGROUP_RUN_PROG_GETSOCKOPT(sock->sk, level, optname, | |
2151 | optval, optlen, | |
2152 | max_optlen, err); | |
6cb153ca BL |
2153 | out_put: |
2154 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2155 | } |
2156 | return err; | |
2157 | } | |
2158 | ||
13a2d70e DB |
2159 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
2160 | char __user *, optval, int __user *, optlen) | |
2161 | { | |
2162 | return __sys_getsockopt(fd, level, optname, optval, optlen); | |
2163 | } | |
2164 | ||
1da177e4 LT |
2165 | /* |
2166 | * Shutdown a socket. | |
2167 | */ | |
2168 | ||
005a1aea | 2169 | int __sys_shutdown(int fd, int how) |
1da177e4 | 2170 | { |
6cb153ca | 2171 | int err, fput_needed; |
1da177e4 LT |
2172 | struct socket *sock; |
2173 | ||
89bddce5 SH |
2174 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2175 | if (sock != NULL) { | |
1da177e4 | 2176 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
2177 | if (!err) |
2178 | err = sock->ops->shutdown(sock, how); | |
2179 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2180 | } |
2181 | return err; | |
2182 | } | |
2183 | ||
005a1aea DB |
2184 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
2185 | { | |
2186 | return __sys_shutdown(fd, how); | |
2187 | } | |
2188 | ||
89bddce5 | 2189 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
2190 | * fields which are the same type (int / unsigned) on our platforms. |
2191 | */ | |
2192 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
2193 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
2194 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
2195 | ||
c71d8ebe TH |
2196 | struct used_address { |
2197 | struct sockaddr_storage name; | |
2198 | unsigned int name_len; | |
2199 | }; | |
2200 | ||
da184284 AV |
2201 | static int copy_msghdr_from_user(struct msghdr *kmsg, |
2202 | struct user_msghdr __user *umsg, | |
2203 | struct sockaddr __user **save_addr, | |
2204 | struct iovec **iov) | |
1661bf36 | 2205 | { |
ffb07550 | 2206 | struct user_msghdr msg; |
08adb7da AV |
2207 | ssize_t err; |
2208 | ||
ffb07550 | 2209 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 2210 | return -EFAULT; |
dbb490b9 | 2211 | |
864d9664 | 2212 | kmsg->msg_control = (void __force *)msg.msg_control; |
ffb07550 AV |
2213 | kmsg->msg_controllen = msg.msg_controllen; |
2214 | kmsg->msg_flags = msg.msg_flags; | |
2215 | ||
2216 | kmsg->msg_namelen = msg.msg_namelen; | |
2217 | if (!msg.msg_name) | |
6a2a2b3a AS |
2218 | kmsg->msg_namelen = 0; |
2219 | ||
dbb490b9 ML |
2220 | if (kmsg->msg_namelen < 0) |
2221 | return -EINVAL; | |
2222 | ||
1661bf36 | 2223 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 2224 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
2225 | |
2226 | if (save_addr) | |
ffb07550 | 2227 | *save_addr = msg.msg_name; |
08adb7da | 2228 | |
ffb07550 | 2229 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 2230 | if (!save_addr) { |
864d9664 PA |
2231 | err = move_addr_to_kernel(msg.msg_name, |
2232 | kmsg->msg_namelen, | |
08adb7da AV |
2233 | kmsg->msg_name); |
2234 | if (err < 0) | |
2235 | return err; | |
2236 | } | |
2237 | } else { | |
2238 | kmsg->msg_name = NULL; | |
2239 | kmsg->msg_namelen = 0; | |
2240 | } | |
2241 | ||
ffb07550 | 2242 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
2243 | return -EMSGSIZE; |
2244 | ||
0345f931 | 2245 | kmsg->msg_iocb = NULL; |
2246 | ||
87e5e6da | 2247 | err = import_iovec(save_addr ? READ : WRITE, |
ffb07550 | 2248 | msg.msg_iov, msg.msg_iovlen, |
da184284 | 2249 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
87e5e6da | 2250 | return err < 0 ? err : 0; |
1661bf36 DC |
2251 | } |
2252 | ||
666547ff | 2253 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2254 | struct msghdr *msg_sys, unsigned int flags, |
28a94d8f TH |
2255 | struct used_address *used_address, |
2256 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 2257 | { |
89bddce5 SH |
2258 | struct compat_msghdr __user *msg_compat = |
2259 | (struct compat_msghdr __user *)msg; | |
230b1839 | 2260 | struct sockaddr_storage address; |
1da177e4 | 2261 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 2262 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 2263 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 2264 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 2265 | unsigned char *ctl_buf = ctl; |
d8725c86 | 2266 | int ctl_len; |
08adb7da | 2267 | ssize_t err; |
89bddce5 | 2268 | |
08adb7da | 2269 | msg_sys->msg_name = &address; |
1da177e4 | 2270 | |
08449320 | 2271 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2272 | err = get_compat_msghdr(msg_sys, msg_compat, NULL, &iov); |
08449320 | 2273 | else |
08adb7da | 2274 | err = copy_msghdr_from_user(msg_sys, msg, NULL, &iov); |
89bddce5 | 2275 | if (err < 0) |
da184284 | 2276 | return err; |
1da177e4 LT |
2277 | |
2278 | err = -ENOBUFS; | |
2279 | ||
228e548e | 2280 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 2281 | goto out_freeiov; |
28a94d8f | 2282 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2283 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2284 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2285 | err = |
228e548e | 2286 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2287 | sizeof(ctl)); |
1da177e4 LT |
2288 | if (err) |
2289 | goto out_freeiov; | |
228e548e AB |
2290 | ctl_buf = msg_sys->msg_control; |
2291 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2292 | } else if (ctl_len) { |
ac4340fc DM |
2293 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2294 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2295 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2296 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2297 | if (ctl_buf == NULL) |
1da177e4 LT |
2298 | goto out_freeiov; |
2299 | } | |
2300 | err = -EFAULT; | |
2301 | /* | |
228e548e | 2302 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2303 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2304 | * checking falls down on this. | |
2305 | */ | |
fb8621bb | 2306 | if (copy_from_user(ctl_buf, |
228e548e | 2307 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2308 | ctl_len)) |
1da177e4 | 2309 | goto out_freectl; |
228e548e | 2310 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2311 | } |
228e548e | 2312 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2313 | |
2314 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2315 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2316 | /* |
2317 | * If this is sendmmsg() and current destination address is same as | |
2318 | * previously succeeded address, omit asking LSM's decision. | |
2319 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2320 | * destination address never matches. | |
2321 | */ | |
bc909d9d MD |
2322 | if (used_address && msg_sys->msg_name && |
2323 | used_address->name_len == msg_sys->msg_namelen && | |
2324 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2325 | used_address->name_len)) { |
d8725c86 | 2326 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2327 | goto out_freectl; |
2328 | } | |
d8725c86 | 2329 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2330 | /* |
2331 | * If this is sendmmsg() and sending to current destination address was | |
2332 | * successful, remember it. | |
2333 | */ | |
2334 | if (used_address && err >= 0) { | |
2335 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2336 | if (msg_sys->msg_name) |
2337 | memcpy(&used_address->name, msg_sys->msg_name, | |
2338 | used_address->name_len); | |
c71d8ebe | 2339 | } |
1da177e4 LT |
2340 | |
2341 | out_freectl: | |
89bddce5 | 2342 | if (ctl_buf != ctl) |
1da177e4 LT |
2343 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
2344 | out_freeiov: | |
da184284 | 2345 | kfree(iov); |
228e548e AB |
2346 | return err; |
2347 | } | |
2348 | ||
2349 | /* | |
2350 | * BSD sendmsg interface | |
2351 | */ | |
0fa03c62 JA |
2352 | long __sys_sendmsg_sock(struct socket *sock, struct user_msghdr __user *msg, |
2353 | unsigned int flags) | |
2354 | { | |
2355 | struct msghdr msg_sys; | |
2356 | ||
2357 | return ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); | |
2358 | } | |
228e548e | 2359 | |
e1834a32 DB |
2360 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2361 | bool forbid_cmsg_compat) | |
228e548e AB |
2362 | { |
2363 | int fput_needed, err; | |
2364 | struct msghdr msg_sys; | |
1be374a0 AL |
2365 | struct socket *sock; |
2366 | ||
e1834a32 DB |
2367 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2368 | return -EINVAL; | |
2369 | ||
1be374a0 | 2370 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2371 | if (!sock) |
2372 | goto out; | |
2373 | ||
28a94d8f | 2374 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2375 | |
6cb153ca | 2376 | fput_light(sock->file, fput_needed); |
89bddce5 | 2377 | out: |
1da177e4 LT |
2378 | return err; |
2379 | } | |
2380 | ||
666547ff | 2381 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 | 2382 | { |
e1834a32 | 2383 | return __sys_sendmsg(fd, msg, flags, true); |
a7526eb5 AL |
2384 | } |
2385 | ||
228e548e AB |
2386 | /* |
2387 | * Linux sendmmsg interface | |
2388 | */ | |
2389 | ||
2390 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
e1834a32 | 2391 | unsigned int flags, bool forbid_cmsg_compat) |
228e548e AB |
2392 | { |
2393 | int fput_needed, err, datagrams; | |
2394 | struct socket *sock; | |
2395 | struct mmsghdr __user *entry; | |
2396 | struct compat_mmsghdr __user *compat_entry; | |
2397 | struct msghdr msg_sys; | |
c71d8ebe | 2398 | struct used_address used_address; |
f092276d | 2399 | unsigned int oflags = flags; |
228e548e | 2400 | |
e1834a32 DB |
2401 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2402 | return -EINVAL; | |
2403 | ||
98382f41 AB |
2404 | if (vlen > UIO_MAXIOV) |
2405 | vlen = UIO_MAXIOV; | |
228e548e AB |
2406 | |
2407 | datagrams = 0; | |
2408 | ||
2409 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2410 | if (!sock) | |
2411 | return err; | |
2412 | ||
c71d8ebe | 2413 | used_address.name_len = UINT_MAX; |
228e548e AB |
2414 | entry = mmsg; |
2415 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2416 | err = 0; |
f092276d | 2417 | flags |= MSG_BATCH; |
228e548e AB |
2418 | |
2419 | while (datagrams < vlen) { | |
f092276d TH |
2420 | if (datagrams == vlen - 1) |
2421 | flags = oflags; | |
2422 | ||
228e548e | 2423 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2424 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2425 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2426 | if (err < 0) |
2427 | break; | |
2428 | err = __put_user(err, &compat_entry->msg_len); | |
2429 | ++compat_entry; | |
2430 | } else { | |
a7526eb5 | 2431 | err = ___sys_sendmsg(sock, |
666547ff | 2432 | (struct user_msghdr __user *)entry, |
28a94d8f | 2433 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2434 | if (err < 0) |
2435 | break; | |
2436 | err = put_user(err, &entry->msg_len); | |
2437 | ++entry; | |
2438 | } | |
2439 | ||
2440 | if (err) | |
2441 | break; | |
2442 | ++datagrams; | |
3023898b SHY |
2443 | if (msg_data_left(&msg_sys)) |
2444 | break; | |
a78cb84c | 2445 | cond_resched(); |
228e548e AB |
2446 | } |
2447 | ||
228e548e AB |
2448 | fput_light(sock->file, fput_needed); |
2449 | ||
728ffb86 AB |
2450 | /* We only return an error if no datagrams were able to be sent */ |
2451 | if (datagrams != 0) | |
228e548e AB |
2452 | return datagrams; |
2453 | ||
228e548e AB |
2454 | return err; |
2455 | } | |
2456 | ||
2457 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2458 | unsigned int, vlen, unsigned int, flags) | |
2459 | { | |
e1834a32 | 2460 | return __sys_sendmmsg(fd, mmsg, vlen, flags, true); |
228e548e AB |
2461 | } |
2462 | ||
666547ff | 2463 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2464 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2465 | { |
89bddce5 SH |
2466 | struct compat_msghdr __user *msg_compat = |
2467 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2468 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2469 | struct iovec *iov = iovstack; |
1da177e4 | 2470 | unsigned long cmsg_ptr; |
2da62906 | 2471 | int len; |
08adb7da | 2472 | ssize_t err; |
1da177e4 LT |
2473 | |
2474 | /* kernel mode address */ | |
230b1839 | 2475 | struct sockaddr_storage addr; |
1da177e4 LT |
2476 | |
2477 | /* user mode address pointers */ | |
2478 | struct sockaddr __user *uaddr; | |
08adb7da | 2479 | int __user *uaddr_len = COMPAT_NAMELEN(msg); |
89bddce5 | 2480 | |
08adb7da | 2481 | msg_sys->msg_name = &addr; |
1da177e4 | 2482 | |
f3d33426 | 2483 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2484 | err = get_compat_msghdr(msg_sys, msg_compat, &uaddr, &iov); |
f3d33426 | 2485 | else |
08adb7da | 2486 | err = copy_msghdr_from_user(msg_sys, msg, &uaddr, &iov); |
1da177e4 | 2487 | if (err < 0) |
da184284 | 2488 | return err; |
1da177e4 | 2489 | |
a2e27255 ACM |
2490 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2491 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2492 | |
f3d33426 HFS |
2493 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2494 | msg_sys->msg_namelen = 0; | |
2495 | ||
1da177e4 LT |
2496 | if (sock->file->f_flags & O_NONBLOCK) |
2497 | flags |= MSG_DONTWAIT; | |
2da62906 | 2498 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, flags); |
1da177e4 LT |
2499 | if (err < 0) |
2500 | goto out_freeiov; | |
2501 | len = err; | |
2502 | ||
2503 | if (uaddr != NULL) { | |
43db362d | 2504 | err = move_addr_to_user(&addr, |
a2e27255 | 2505 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2506 | uaddr_len); |
1da177e4 LT |
2507 | if (err < 0) |
2508 | goto out_freeiov; | |
2509 | } | |
a2e27255 | 2510 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2511 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2512 | if (err) |
2513 | goto out_freeiov; | |
2514 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2515 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2516 | &msg_compat->msg_controllen); |
2517 | else | |
a2e27255 | 2518 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2519 | &msg->msg_controllen); |
2520 | if (err) | |
2521 | goto out_freeiov; | |
2522 | err = len; | |
2523 | ||
2524 | out_freeiov: | |
da184284 | 2525 | kfree(iov); |
a2e27255 ACM |
2526 | return err; |
2527 | } | |
2528 | ||
2529 | /* | |
2530 | * BSD recvmsg interface | |
2531 | */ | |
2532 | ||
aa1fa28f JA |
2533 | long __sys_recvmsg_sock(struct socket *sock, struct user_msghdr __user *msg, |
2534 | unsigned int flags) | |
2535 | { | |
2536 | struct msghdr msg_sys; | |
2537 | ||
2538 | return ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); | |
2539 | } | |
2540 | ||
e1834a32 DB |
2541 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2542 | bool forbid_cmsg_compat) | |
a2e27255 ACM |
2543 | { |
2544 | int fput_needed, err; | |
2545 | struct msghdr msg_sys; | |
1be374a0 AL |
2546 | struct socket *sock; |
2547 | ||
e1834a32 DB |
2548 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2549 | return -EINVAL; | |
2550 | ||
1be374a0 | 2551 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2552 | if (!sock) |
2553 | goto out; | |
2554 | ||
a7526eb5 | 2555 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2556 | |
6cb153ca | 2557 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2558 | out: |
2559 | return err; | |
2560 | } | |
2561 | ||
666547ff | 2562 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2563 | unsigned int, flags) |
2564 | { | |
e1834a32 | 2565 | return __sys_recvmsg(fd, msg, flags, true); |
a7526eb5 AL |
2566 | } |
2567 | ||
a2e27255 ACM |
2568 | /* |
2569 | * Linux recvmmsg interface | |
2570 | */ | |
2571 | ||
e11d4284 AB |
2572 | static int do_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2573 | unsigned int vlen, unsigned int flags, | |
2574 | struct timespec64 *timeout) | |
a2e27255 ACM |
2575 | { |
2576 | int fput_needed, err, datagrams; | |
2577 | struct socket *sock; | |
2578 | struct mmsghdr __user *entry; | |
d7256d0e | 2579 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2580 | struct msghdr msg_sys; |
766b9f92 DD |
2581 | struct timespec64 end_time; |
2582 | struct timespec64 timeout64; | |
a2e27255 ACM |
2583 | |
2584 | if (timeout && | |
2585 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2586 | timeout->tv_nsec)) | |
2587 | return -EINVAL; | |
2588 | ||
2589 | datagrams = 0; | |
2590 | ||
2591 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2592 | if (!sock) | |
2593 | return err; | |
2594 | ||
7797dc41 SHY |
2595 | if (likely(!(flags & MSG_ERRQUEUE))) { |
2596 | err = sock_error(sock->sk); | |
2597 | if (err) { | |
2598 | datagrams = err; | |
2599 | goto out_put; | |
2600 | } | |
e623a9e9 | 2601 | } |
a2e27255 ACM |
2602 | |
2603 | entry = mmsg; | |
d7256d0e | 2604 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2605 | |
2606 | while (datagrams < vlen) { | |
2607 | /* | |
2608 | * No need to ask LSM for more than the first datagram. | |
2609 | */ | |
d7256d0e | 2610 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2611 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2612 | &msg_sys, flags & ~MSG_WAITFORONE, |
2613 | datagrams); | |
d7256d0e JMG |
2614 | if (err < 0) |
2615 | break; | |
2616 | err = __put_user(err, &compat_entry->msg_len); | |
2617 | ++compat_entry; | |
2618 | } else { | |
a7526eb5 | 2619 | err = ___sys_recvmsg(sock, |
666547ff | 2620 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2621 | &msg_sys, flags & ~MSG_WAITFORONE, |
2622 | datagrams); | |
d7256d0e JMG |
2623 | if (err < 0) |
2624 | break; | |
2625 | err = put_user(err, &entry->msg_len); | |
2626 | ++entry; | |
2627 | } | |
2628 | ||
a2e27255 ACM |
2629 | if (err) |
2630 | break; | |
a2e27255 ACM |
2631 | ++datagrams; |
2632 | ||
71c5c159 BB |
2633 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2634 | if (flags & MSG_WAITFORONE) | |
2635 | flags |= MSG_DONTWAIT; | |
2636 | ||
a2e27255 | 2637 | if (timeout) { |
766b9f92 | 2638 | ktime_get_ts64(&timeout64); |
c2e6c856 | 2639 | *timeout = timespec64_sub(end_time, timeout64); |
a2e27255 ACM |
2640 | if (timeout->tv_sec < 0) { |
2641 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2642 | break; | |
2643 | } | |
2644 | ||
2645 | /* Timeout, return less than vlen datagrams */ | |
2646 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2647 | break; | |
2648 | } | |
2649 | ||
2650 | /* Out of band data, return right away */ | |
2651 | if (msg_sys.msg_flags & MSG_OOB) | |
2652 | break; | |
a78cb84c | 2653 | cond_resched(); |
a2e27255 ACM |
2654 | } |
2655 | ||
a2e27255 | 2656 | if (err == 0) |
34b88a68 ACM |
2657 | goto out_put; |
2658 | ||
2659 | if (datagrams == 0) { | |
2660 | datagrams = err; | |
2661 | goto out_put; | |
2662 | } | |
a2e27255 | 2663 | |
34b88a68 ACM |
2664 | /* |
2665 | * We may return less entries than requested (vlen) if the | |
2666 | * sock is non block and there aren't enough datagrams... | |
2667 | */ | |
2668 | if (err != -EAGAIN) { | |
a2e27255 | 2669 | /* |
34b88a68 ACM |
2670 | * ... or if recvmsg returns an error after we |
2671 | * received some datagrams, where we record the | |
2672 | * error to return on the next call or if the | |
2673 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2674 | */ |
34b88a68 | 2675 | sock->sk->sk_err = -err; |
a2e27255 | 2676 | } |
34b88a68 ACM |
2677 | out_put: |
2678 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2679 | |
34b88a68 | 2680 | return datagrams; |
a2e27255 ACM |
2681 | } |
2682 | ||
e11d4284 AB |
2683 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2684 | unsigned int vlen, unsigned int flags, | |
2685 | struct __kernel_timespec __user *timeout, | |
2686 | struct old_timespec32 __user *timeout32) | |
a2e27255 ACM |
2687 | { |
2688 | int datagrams; | |
c2e6c856 | 2689 | struct timespec64 timeout_sys; |
a2e27255 | 2690 | |
e11d4284 AB |
2691 | if (timeout && get_timespec64(&timeout_sys, timeout)) |
2692 | return -EFAULT; | |
a2e27255 | 2693 | |
e11d4284 | 2694 | if (timeout32 && get_old_timespec32(&timeout_sys, timeout32)) |
a2e27255 ACM |
2695 | return -EFAULT; |
2696 | ||
e11d4284 AB |
2697 | if (!timeout && !timeout32) |
2698 | return do_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2699 | ||
2700 | datagrams = do_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
a2e27255 | 2701 | |
e11d4284 AB |
2702 | if (datagrams <= 0) |
2703 | return datagrams; | |
2704 | ||
2705 | if (timeout && put_timespec64(&timeout_sys, timeout)) | |
2706 | datagrams = -EFAULT; | |
2707 | ||
2708 | if (timeout32 && put_old_timespec32(&timeout_sys, timeout32)) | |
a2e27255 ACM |
2709 | datagrams = -EFAULT; |
2710 | ||
2711 | return datagrams; | |
2712 | } | |
2713 | ||
1255e269 DB |
2714 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, |
2715 | unsigned int, vlen, unsigned int, flags, | |
c2e6c856 | 2716 | struct __kernel_timespec __user *, timeout) |
1255e269 | 2717 | { |
e11d4284 AB |
2718 | if (flags & MSG_CMSG_COMPAT) |
2719 | return -EINVAL; | |
2720 | ||
2721 | return __sys_recvmmsg(fd, mmsg, vlen, flags, timeout, NULL); | |
2722 | } | |
2723 | ||
2724 | #ifdef CONFIG_COMPAT_32BIT_TIME | |
2725 | SYSCALL_DEFINE5(recvmmsg_time32, int, fd, struct mmsghdr __user *, mmsg, | |
2726 | unsigned int, vlen, unsigned int, flags, | |
2727 | struct old_timespec32 __user *, timeout) | |
2728 | { | |
2729 | if (flags & MSG_CMSG_COMPAT) | |
2730 | return -EINVAL; | |
2731 | ||
2732 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL, timeout); | |
1255e269 | 2733 | } |
e11d4284 | 2734 | #endif |
1255e269 | 2735 | |
a2e27255 | 2736 | #ifdef __ARCH_WANT_SYS_SOCKETCALL |
1da177e4 LT |
2737 | /* Argument list sizes for sys_socketcall */ |
2738 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2739 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2740 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2741 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2742 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2743 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2744 | }; |
2745 | ||
1da177e4 LT |
2746 | #undef AL |
2747 | ||
2748 | /* | |
89bddce5 | 2749 | * System call vectors. |
1da177e4 LT |
2750 | * |
2751 | * Argument checking cleaned up. Saved 20% in size. | |
2752 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2753 | * it is set by the callees. |
1da177e4 LT |
2754 | */ |
2755 | ||
3e0fa65f | 2756 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2757 | { |
2950fa9d | 2758 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2759 | unsigned long a0, a1; |
1da177e4 | 2760 | int err; |
47379052 | 2761 | unsigned int len; |
1da177e4 | 2762 | |
228e548e | 2763 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 | 2764 | return -EINVAL; |
c8e8cd57 | 2765 | call = array_index_nospec(call, SYS_SENDMMSG + 1); |
1da177e4 | 2766 | |
47379052 AV |
2767 | len = nargs[call]; |
2768 | if (len > sizeof(a)) | |
2769 | return -EINVAL; | |
2770 | ||
1da177e4 | 2771 | /* copy_from_user should be SMP safe. */ |
47379052 | 2772 | if (copy_from_user(a, args, len)) |
1da177e4 | 2773 | return -EFAULT; |
3ec3b2fb | 2774 | |
2950fa9d CG |
2775 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2776 | if (err) | |
2777 | return err; | |
3ec3b2fb | 2778 | |
89bddce5 SH |
2779 | a0 = a[0]; |
2780 | a1 = a[1]; | |
2781 | ||
2782 | switch (call) { | |
2783 | case SYS_SOCKET: | |
9d6a15c3 | 2784 | err = __sys_socket(a0, a1, a[2]); |
89bddce5 SH |
2785 | break; |
2786 | case SYS_BIND: | |
a87d35d8 | 2787 | err = __sys_bind(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2788 | break; |
2789 | case SYS_CONNECT: | |
1387c2c2 | 2790 | err = __sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2791 | break; |
2792 | case SYS_LISTEN: | |
25e290ee | 2793 | err = __sys_listen(a0, a1); |
89bddce5 SH |
2794 | break; |
2795 | case SYS_ACCEPT: | |
4541e805 DB |
2796 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2797 | (int __user *)a[2], 0); | |
89bddce5 SH |
2798 | break; |
2799 | case SYS_GETSOCKNAME: | |
2800 | err = | |
8882a107 DB |
2801 | __sys_getsockname(a0, (struct sockaddr __user *)a1, |
2802 | (int __user *)a[2]); | |
89bddce5 SH |
2803 | break; |
2804 | case SYS_GETPEERNAME: | |
2805 | err = | |
b21c8f83 DB |
2806 | __sys_getpeername(a0, (struct sockaddr __user *)a1, |
2807 | (int __user *)a[2]); | |
89bddce5 SH |
2808 | break; |
2809 | case SYS_SOCKETPAIR: | |
6debc8d8 | 2810 | err = __sys_socketpair(a0, a1, a[2], (int __user *)a[3]); |
89bddce5 SH |
2811 | break; |
2812 | case SYS_SEND: | |
f3bf896b DB |
2813 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2814 | NULL, 0); | |
89bddce5 SH |
2815 | break; |
2816 | case SYS_SENDTO: | |
211b634b DB |
2817 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2818 | (struct sockaddr __user *)a[4], a[5]); | |
89bddce5 SH |
2819 | break; |
2820 | case SYS_RECV: | |
d27e9afc DB |
2821 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2822 | NULL, NULL); | |
89bddce5 SH |
2823 | break; |
2824 | case SYS_RECVFROM: | |
7a09e1eb DB |
2825 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2826 | (struct sockaddr __user *)a[4], | |
2827 | (int __user *)a[5]); | |
89bddce5 SH |
2828 | break; |
2829 | case SYS_SHUTDOWN: | |
005a1aea | 2830 | err = __sys_shutdown(a0, a1); |
89bddce5 SH |
2831 | break; |
2832 | case SYS_SETSOCKOPT: | |
cc36dca0 DB |
2833 | err = __sys_setsockopt(a0, a1, a[2], (char __user *)a[3], |
2834 | a[4]); | |
89bddce5 SH |
2835 | break; |
2836 | case SYS_GETSOCKOPT: | |
2837 | err = | |
13a2d70e DB |
2838 | __sys_getsockopt(a0, a1, a[2], (char __user *)a[3], |
2839 | (int __user *)a[4]); | |
89bddce5 SH |
2840 | break; |
2841 | case SYS_SENDMSG: | |
e1834a32 DB |
2842 | err = __sys_sendmsg(a0, (struct user_msghdr __user *)a1, |
2843 | a[2], true); | |
89bddce5 | 2844 | break; |
228e548e | 2845 | case SYS_SENDMMSG: |
e1834a32 DB |
2846 | err = __sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
2847 | a[3], true); | |
228e548e | 2848 | break; |
89bddce5 | 2849 | case SYS_RECVMSG: |
e1834a32 DB |
2850 | err = __sys_recvmsg(a0, (struct user_msghdr __user *)a1, |
2851 | a[2], true); | |
89bddce5 | 2852 | break; |
a2e27255 | 2853 | case SYS_RECVMMSG: |
e11d4284 AB |
2854 | if (IS_ENABLED(CONFIG_64BIT) || !IS_ENABLED(CONFIG_64BIT_TIME)) |
2855 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, | |
2856 | a[2], a[3], | |
2857 | (struct __kernel_timespec __user *)a[4], | |
2858 | NULL); | |
2859 | else | |
2860 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, | |
2861 | a[2], a[3], NULL, | |
2862 | (struct old_timespec32 __user *)a[4]); | |
a2e27255 | 2863 | break; |
de11defe | 2864 | case SYS_ACCEPT4: |
4541e805 DB |
2865 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2866 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2867 | break; |
89bddce5 SH |
2868 | default: |
2869 | err = -EINVAL; | |
2870 | break; | |
1da177e4 LT |
2871 | } |
2872 | return err; | |
2873 | } | |
2874 | ||
89bddce5 | 2875 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2876 | |
55737fda SH |
2877 | /** |
2878 | * sock_register - add a socket protocol handler | |
2879 | * @ops: description of protocol | |
2880 | * | |
1da177e4 LT |
2881 | * This function is called by a protocol handler that wants to |
2882 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2883 | * socket interface. The value ops->family corresponds to the |
55737fda | 2884 | * socket system call protocol family. |
1da177e4 | 2885 | */ |
f0fd27d4 | 2886 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2887 | { |
2888 | int err; | |
2889 | ||
2890 | if (ops->family >= NPROTO) { | |
3410f22e | 2891 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2892 | return -ENOBUFS; |
2893 | } | |
55737fda SH |
2894 | |
2895 | spin_lock(&net_family_lock); | |
190683a9 ED |
2896 | if (rcu_dereference_protected(net_families[ops->family], |
2897 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2898 | err = -EEXIST; |
2899 | else { | |
cf778b00 | 2900 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2901 | err = 0; |
2902 | } | |
55737fda SH |
2903 | spin_unlock(&net_family_lock); |
2904 | ||
3410f22e | 2905 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2906 | return err; |
2907 | } | |
c6d409cf | 2908 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2909 | |
55737fda SH |
2910 | /** |
2911 | * sock_unregister - remove a protocol handler | |
2912 | * @family: protocol family to remove | |
2913 | * | |
1da177e4 LT |
2914 | * This function is called by a protocol handler that wants to |
2915 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2916 | * new socket creation. |
2917 | * | |
2918 | * If protocol handler is a module, then it can use module reference | |
2919 | * counts to protect against new references. If protocol handler is not | |
2920 | * a module then it needs to provide its own protection in | |
2921 | * the ops->create routine. | |
1da177e4 | 2922 | */ |
f0fd27d4 | 2923 | void sock_unregister(int family) |
1da177e4 | 2924 | { |
f0fd27d4 | 2925 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2926 | |
55737fda | 2927 | spin_lock(&net_family_lock); |
a9b3cd7f | 2928 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2929 | spin_unlock(&net_family_lock); |
2930 | ||
2931 | synchronize_rcu(); | |
2932 | ||
3410f22e | 2933 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2934 | } |
c6d409cf | 2935 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2936 | |
bf2ae2e4 XL |
2937 | bool sock_is_registered(int family) |
2938 | { | |
66b51b0a | 2939 | return family < NPROTO && rcu_access_pointer(net_families[family]); |
bf2ae2e4 XL |
2940 | } |
2941 | ||
77d76ea3 | 2942 | static int __init sock_init(void) |
1da177e4 | 2943 | { |
b3e19d92 | 2944 | int err; |
2ca794e5 EB |
2945 | /* |
2946 | * Initialize the network sysctl infrastructure. | |
2947 | */ | |
2948 | err = net_sysctl_init(); | |
2949 | if (err) | |
2950 | goto out; | |
b3e19d92 | 2951 | |
1da177e4 | 2952 | /* |
89bddce5 | 2953 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2954 | */ |
2955 | skb_init(); | |
1da177e4 LT |
2956 | |
2957 | /* | |
89bddce5 | 2958 | * Initialize the protocols module. |
1da177e4 LT |
2959 | */ |
2960 | ||
2961 | init_inodecache(); | |
b3e19d92 NP |
2962 | |
2963 | err = register_filesystem(&sock_fs_type); | |
2964 | if (err) | |
2965 | goto out_fs; | |
1da177e4 | 2966 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2967 | if (IS_ERR(sock_mnt)) { |
2968 | err = PTR_ERR(sock_mnt); | |
2969 | goto out_mount; | |
2970 | } | |
77d76ea3 AK |
2971 | |
2972 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2973 | */ |
2974 | ||
2975 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
2976 | err = netfilter_init(); |
2977 | if (err) | |
2978 | goto out; | |
1da177e4 | 2979 | #endif |
cbeb321a | 2980 | |
408eccce | 2981 | ptp_classifier_init(); |
c1f19b51 | 2982 | |
b3e19d92 NP |
2983 | out: |
2984 | return err; | |
2985 | ||
2986 | out_mount: | |
2987 | unregister_filesystem(&sock_fs_type); | |
2988 | out_fs: | |
2989 | goto out; | |
1da177e4 LT |
2990 | } |
2991 | ||
77d76ea3 AK |
2992 | core_initcall(sock_init); /* early initcall */ |
2993 | ||
1da177e4 LT |
2994 | #ifdef CONFIG_PROC_FS |
2995 | void socket_seq_show(struct seq_file *seq) | |
2996 | { | |
648845ab TZ |
2997 | seq_printf(seq, "sockets: used %d\n", |
2998 | sock_inuse_get(seq->private)); | |
1da177e4 | 2999 | } |
89bddce5 | 3000 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 3001 | |
89bbfc95 | 3002 | #ifdef CONFIG_COMPAT |
36fd633e | 3003 | static int compat_dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 3004 | { |
6b96018b | 3005 | struct compat_ifconf ifc32; |
7a229387 | 3006 | struct ifconf ifc; |
7a229387 AB |
3007 | int err; |
3008 | ||
6b96018b | 3009 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
3010 | return -EFAULT; |
3011 | ||
36fd633e AV |
3012 | ifc.ifc_len = ifc32.ifc_len; |
3013 | ifc.ifc_req = compat_ptr(ifc32.ifcbuf); | |
7a229387 | 3014 | |
36fd633e AV |
3015 | rtnl_lock(); |
3016 | err = dev_ifconf(net, &ifc, sizeof(struct compat_ifreq)); | |
3017 | rtnl_unlock(); | |
7a229387 AB |
3018 | if (err) |
3019 | return err; | |
3020 | ||
36fd633e | 3021 | ifc32.ifc_len = ifc.ifc_len; |
6b96018b | 3022 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
3023 | return -EFAULT; |
3024 | ||
3025 | return 0; | |
3026 | } | |
3027 | ||
6b96018b | 3028 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 3029 | { |
3a7da39d BH |
3030 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
3031 | bool convert_in = false, convert_out = false; | |
44c02a2c AV |
3032 | size_t buf_size = 0; |
3033 | struct ethtool_rxnfc __user *rxnfc = NULL; | |
3034 | struct ifreq ifr; | |
3a7da39d BH |
3035 | u32 rule_cnt = 0, actual_rule_cnt; |
3036 | u32 ethcmd; | |
7a229387 | 3037 | u32 data; |
3a7da39d | 3038 | int ret; |
7a229387 | 3039 | |
3a7da39d BH |
3040 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
3041 | return -EFAULT; | |
7a229387 | 3042 | |
3a7da39d BH |
3043 | compat_rxnfc = compat_ptr(data); |
3044 | ||
3045 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
3046 | return -EFAULT; |
3047 | ||
3a7da39d BH |
3048 | /* Most ethtool structures are defined without padding. |
3049 | * Unfortunately struct ethtool_rxnfc is an exception. | |
3050 | */ | |
3051 | switch (ethcmd) { | |
3052 | default: | |
3053 | break; | |
3054 | case ETHTOOL_GRXCLSRLALL: | |
3055 | /* Buffer size is variable */ | |
3056 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
3057 | return -EFAULT; | |
3058 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
3059 | return -ENOMEM; | |
3060 | buf_size += rule_cnt * sizeof(u32); | |
3061 | /* fall through */ | |
3062 | case ETHTOOL_GRXRINGS: | |
3063 | case ETHTOOL_GRXCLSRLCNT: | |
3064 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 3065 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
3066 | convert_out = true; |
3067 | /* fall through */ | |
3068 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
3069 | buf_size += sizeof(struct ethtool_rxnfc); |
3070 | convert_in = true; | |
44c02a2c | 3071 | rxnfc = compat_alloc_user_space(buf_size); |
3a7da39d BH |
3072 | break; |
3073 | } | |
3074 | ||
44c02a2c | 3075 | if (copy_from_user(&ifr.ifr_name, &ifr32->ifr_name, IFNAMSIZ)) |
7a229387 AB |
3076 | return -EFAULT; |
3077 | ||
44c02a2c | 3078 | ifr.ifr_data = convert_in ? rxnfc : (void __user *)compat_rxnfc; |
7a229387 | 3079 | |
3a7da39d | 3080 | if (convert_in) { |
127fe533 | 3081 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
3082 | * fs.ring_cookie and at the end of fs, but nowhere else. |
3083 | */ | |
127fe533 AD |
3084 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
3085 | sizeof(compat_rxnfc->fs.m_ext) != | |
3086 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
3087 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
3088 | BUILD_BUG_ON( |
3089 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
3090 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
3091 | offsetof(struct ethtool_rxnfc, fs.location) - | |
3092 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
3093 | ||
3094 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
3095 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
3096 | (void __user *)rxnfc) || | |
3a7da39d BH |
3097 | copy_in_user(&rxnfc->fs.ring_cookie, |
3098 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 | 3099 | (void __user *)(&rxnfc->fs.location + 1) - |
b6168562 WW |
3100 | (void __user *)&rxnfc->fs.ring_cookie)) |
3101 | return -EFAULT; | |
3102 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3103 | if (put_user(rule_cnt, &rxnfc->rule_cnt)) | |
3104 | return -EFAULT; | |
3105 | } else if (copy_in_user(&rxnfc->rule_cnt, | |
3106 | &compat_rxnfc->rule_cnt, | |
3107 | sizeof(rxnfc->rule_cnt))) | |
3a7da39d BH |
3108 | return -EFAULT; |
3109 | } | |
3110 | ||
44c02a2c | 3111 | ret = dev_ioctl(net, SIOCETHTOOL, &ifr, NULL); |
3a7da39d BH |
3112 | if (ret) |
3113 | return ret; | |
3114 | ||
3115 | if (convert_out) { | |
3116 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
3117 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
3118 | (const void __user *)rxnfc) || | |
3a7da39d BH |
3119 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
3120 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
3121 | (const void __user *)(&rxnfc->fs.location + 1) - |
3122 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
3123 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
3124 | sizeof(rxnfc->rule_cnt))) | |
3125 | return -EFAULT; | |
3126 | ||
3127 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3128 | /* As an optimisation, we only copy the actual | |
3129 | * number of rules that the underlying | |
3130 | * function returned. Since Mallory might | |
3131 | * change the rule count in user memory, we | |
3132 | * check that it is less than the rule count | |
3133 | * originally given (as the user buffer size), | |
3134 | * which has been range-checked. | |
3135 | */ | |
3136 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
3137 | return -EFAULT; | |
3138 | if (actual_rule_cnt < rule_cnt) | |
3139 | rule_cnt = actual_rule_cnt; | |
3140 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
3141 | &rxnfc->rule_locs[0], | |
3142 | rule_cnt * sizeof(u32))) | |
3143 | return -EFAULT; | |
3144 | } | |
3145 | } | |
3146 | ||
3147 | return 0; | |
7a229387 AB |
3148 | } |
3149 | ||
7a50a240 AB |
3150 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
3151 | { | |
7a50a240 | 3152 | compat_uptr_t uptr32; |
44c02a2c AV |
3153 | struct ifreq ifr; |
3154 | void __user *saved; | |
3155 | int err; | |
7a50a240 | 3156 | |
44c02a2c | 3157 | if (copy_from_user(&ifr, uifr32, sizeof(struct compat_ifreq))) |
7a50a240 AB |
3158 | return -EFAULT; |
3159 | ||
3160 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
3161 | return -EFAULT; | |
3162 | ||
44c02a2c AV |
3163 | saved = ifr.ifr_settings.ifs_ifsu.raw_hdlc; |
3164 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = compat_ptr(uptr32); | |
7a229387 | 3165 | |
44c02a2c AV |
3166 | err = dev_ioctl(net, SIOCWANDEV, &ifr, NULL); |
3167 | if (!err) { | |
3168 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = saved; | |
3169 | if (copy_to_user(uifr32, &ifr, sizeof(struct compat_ifreq))) | |
3170 | err = -EFAULT; | |
ccbd6a5a | 3171 | } |
44c02a2c | 3172 | return err; |
7a229387 AB |
3173 | } |
3174 | ||
590d4693 BH |
3175 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
3176 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 3177 | struct compat_ifreq __user *u_ifreq32) |
7a229387 | 3178 | { |
44c02a2c | 3179 | struct ifreq ifreq; |
7a229387 AB |
3180 | u32 data32; |
3181 | ||
44c02a2c | 3182 | if (copy_from_user(ifreq.ifr_name, u_ifreq32->ifr_name, IFNAMSIZ)) |
7a229387 | 3183 | return -EFAULT; |
44c02a2c | 3184 | if (get_user(data32, &u_ifreq32->ifr_data)) |
7a229387 | 3185 | return -EFAULT; |
44c02a2c | 3186 | ifreq.ifr_data = compat_ptr(data32); |
7a229387 | 3187 | |
44c02a2c | 3188 | return dev_ioctl(net, cmd, &ifreq, NULL); |
7a229387 AB |
3189 | } |
3190 | ||
37ac39bd JB |
3191 | static int compat_ifreq_ioctl(struct net *net, struct socket *sock, |
3192 | unsigned int cmd, | |
3193 | struct compat_ifreq __user *uifr32) | |
3194 | { | |
3195 | struct ifreq __user *uifr; | |
3196 | int err; | |
3197 | ||
3198 | /* Handle the fact that while struct ifreq has the same *layout* on | |
3199 | * 32/64 for everything but ifreq::ifru_ifmap and ifreq::ifru_data, | |
3200 | * which are handled elsewhere, it still has different *size* due to | |
3201 | * ifreq::ifru_ifmap (which is 16 bytes on 32 bit, 24 bytes on 64-bit, | |
3202 | * resulting in struct ifreq being 32 and 40 bytes respectively). | |
3203 | * As a result, if the struct happens to be at the end of a page and | |
3204 | * the next page isn't readable/writable, we get a fault. To prevent | |
3205 | * that, copy back and forth to the full size. | |
3206 | */ | |
3207 | ||
3208 | uifr = compat_alloc_user_space(sizeof(*uifr)); | |
3209 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
3210 | return -EFAULT; | |
3211 | ||
3212 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
3213 | ||
3214 | if (!err) { | |
3215 | switch (cmd) { | |
3216 | case SIOCGIFFLAGS: | |
3217 | case SIOCGIFMETRIC: | |
3218 | case SIOCGIFMTU: | |
3219 | case SIOCGIFMEM: | |
3220 | case SIOCGIFHWADDR: | |
3221 | case SIOCGIFINDEX: | |
3222 | case SIOCGIFADDR: | |
3223 | case SIOCGIFBRDADDR: | |
3224 | case SIOCGIFDSTADDR: | |
3225 | case SIOCGIFNETMASK: | |
3226 | case SIOCGIFPFLAGS: | |
3227 | case SIOCGIFTXQLEN: | |
3228 | case SIOCGMIIPHY: | |
3229 | case SIOCGMIIREG: | |
c6c9fee3 | 3230 | case SIOCGIFNAME: |
37ac39bd JB |
3231 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
3232 | err = -EFAULT; | |
3233 | break; | |
3234 | } | |
3235 | } | |
3236 | return err; | |
3237 | } | |
3238 | ||
a2116ed2 AB |
3239 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
3240 | struct compat_ifreq __user *uifr32) | |
3241 | { | |
3242 | struct ifreq ifr; | |
3243 | struct compat_ifmap __user *uifmap32; | |
a2116ed2 AB |
3244 | int err; |
3245 | ||
3246 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
3247 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3248 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3249 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3250 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3251 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3252 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3253 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3254 | if (err) |
3255 | return -EFAULT; | |
3256 | ||
44c02a2c | 3257 | err = dev_ioctl(net, cmd, &ifr, NULL); |
a2116ed2 AB |
3258 | |
3259 | if (cmd == SIOCGIFMAP && !err) { | |
3260 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3261 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3262 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3263 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3264 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3265 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3266 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3267 | if (err) |
3268 | err = -EFAULT; | |
3269 | } | |
3270 | return err; | |
3271 | } | |
3272 | ||
7a229387 | 3273 | struct rtentry32 { |
c6d409cf | 3274 | u32 rt_pad1; |
7a229387 AB |
3275 | struct sockaddr rt_dst; /* target address */ |
3276 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
3277 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
3278 | unsigned short rt_flags; |
3279 | short rt_pad2; | |
3280 | u32 rt_pad3; | |
3281 | unsigned char rt_tos; | |
3282 | unsigned char rt_class; | |
3283 | short rt_pad4; | |
3284 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 3285 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
3286 | u32 rt_mtu; /* per route MTU/Window */ |
3287 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
3288 | unsigned short rt_irtt; /* Initial RTT */ |
3289 | }; | |
3290 | ||
3291 | struct in6_rtmsg32 { | |
3292 | struct in6_addr rtmsg_dst; | |
3293 | struct in6_addr rtmsg_src; | |
3294 | struct in6_addr rtmsg_gateway; | |
3295 | u32 rtmsg_type; | |
3296 | u16 rtmsg_dst_len; | |
3297 | u16 rtmsg_src_len; | |
3298 | u32 rtmsg_metric; | |
3299 | u32 rtmsg_info; | |
3300 | u32 rtmsg_flags; | |
3301 | s32 rtmsg_ifindex; | |
3302 | }; | |
3303 | ||
6b96018b AB |
3304 | static int routing_ioctl(struct net *net, struct socket *sock, |
3305 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3306 | { |
3307 | int ret; | |
3308 | void *r = NULL; | |
3309 | struct in6_rtmsg r6; | |
3310 | struct rtentry r4; | |
3311 | char devname[16]; | |
3312 | u32 rtdev; | |
3313 | mm_segment_t old_fs = get_fs(); | |
3314 | ||
6b96018b AB |
3315 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3316 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3317 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3318 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
3319 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3320 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3321 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3322 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3323 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3324 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3325 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3326 | |
3327 | r = (void *) &r6; | |
3328 | } else { /* ipv4 */ | |
6b96018b | 3329 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3330 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3331 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
3332 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
3333 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3334 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3335 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
3336 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3337 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3338 | if (rtdev) { |
c6d409cf | 3339 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3340 | r4.rt_dev = (char __user __force *)devname; |
3341 | devname[15] = 0; | |
7a229387 AB |
3342 | } else |
3343 | r4.rt_dev = NULL; | |
3344 | ||
3345 | r = (void *) &r4; | |
3346 | } | |
3347 | ||
3348 | if (ret) { | |
3349 | ret = -EFAULT; | |
3350 | goto out; | |
3351 | } | |
3352 | ||
c6d409cf | 3353 | set_fs(KERNEL_DS); |
63ff03ab | 3354 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3355 | set_fs(old_fs); |
7a229387 AB |
3356 | |
3357 | out: | |
7a229387 AB |
3358 | return ret; |
3359 | } | |
3360 | ||
3361 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3362 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3363 | * use compatible ioctls |
7a229387 | 3364 | */ |
6b96018b | 3365 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3366 | { |
6b96018b | 3367 | compat_ulong_t tmp; |
7a229387 | 3368 | |
6b96018b | 3369 | if (get_user(tmp, argp)) |
7a229387 AB |
3370 | return -EFAULT; |
3371 | if (tmp == BRCTL_GET_VERSION) | |
3372 | return BRCTL_VERSION + 1; | |
3373 | return -EINVAL; | |
3374 | } | |
3375 | ||
6b96018b AB |
3376 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3377 | unsigned int cmd, unsigned long arg) | |
3378 | { | |
3379 | void __user *argp = compat_ptr(arg); | |
3380 | struct sock *sk = sock->sk; | |
3381 | struct net *net = sock_net(sk); | |
7a229387 | 3382 | |
6b96018b | 3383 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3384 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3385 | |
3386 | switch (cmd) { | |
3387 | case SIOCSIFBR: | |
3388 | case SIOCGIFBR: | |
3389 | return old_bridge_ioctl(argp); | |
6b96018b | 3390 | case SIOCGIFCONF: |
36fd633e | 3391 | return compat_dev_ifconf(net, argp); |
6b96018b AB |
3392 | case SIOCETHTOOL: |
3393 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3394 | case SIOCWANDEV: |
3395 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3396 | case SIOCGIFMAP: |
3397 | case SIOCSIFMAP: | |
3398 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3399 | case SIOCADDRT: |
3400 | case SIOCDELRT: | |
3401 | return routing_ioctl(net, sock, cmd, argp); | |
0768e170 AB |
3402 | case SIOCGSTAMP_OLD: |
3403 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
3404 | if (!sock->ops->gettstamp) |
3405 | return -ENOIOCTLCMD; | |
0768e170 | 3406 | return sock->ops->gettstamp(sock, argp, cmd == SIOCGSTAMP_OLD, |
c7cbdbf2 AB |
3407 | !COMPAT_USE_64BIT_TIME); |
3408 | ||
590d4693 BH |
3409 | case SIOCBONDSLAVEINFOQUERY: |
3410 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3411 | case SIOCSHWTSTAMP: |
fd468c74 | 3412 | case SIOCGHWTSTAMP: |
590d4693 | 3413 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3414 | |
3415 | case FIOSETOWN: | |
3416 | case SIOCSPGRP: | |
3417 | case FIOGETOWN: | |
3418 | case SIOCGPGRP: | |
3419 | case SIOCBRADDBR: | |
3420 | case SIOCBRDELBR: | |
3421 | case SIOCGIFVLAN: | |
3422 | case SIOCSIFVLAN: | |
3423 | case SIOCADDDLCI: | |
3424 | case SIOCDELDLCI: | |
c62cce2c | 3425 | case SIOCGSKNS: |
0768e170 AB |
3426 | case SIOCGSTAMP_NEW: |
3427 | case SIOCGSTAMPNS_NEW: | |
6b96018b AB |
3428 | return sock_ioctl(file, cmd, arg); |
3429 | ||
3430 | case SIOCGIFFLAGS: | |
3431 | case SIOCSIFFLAGS: | |
3432 | case SIOCGIFMETRIC: | |
3433 | case SIOCSIFMETRIC: | |
3434 | case SIOCGIFMTU: | |
3435 | case SIOCSIFMTU: | |
3436 | case SIOCGIFMEM: | |
3437 | case SIOCSIFMEM: | |
3438 | case SIOCGIFHWADDR: | |
3439 | case SIOCSIFHWADDR: | |
3440 | case SIOCADDMULTI: | |
3441 | case SIOCDELMULTI: | |
3442 | case SIOCGIFINDEX: | |
6b96018b AB |
3443 | case SIOCGIFADDR: |
3444 | case SIOCSIFADDR: | |
3445 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3446 | case SIOCDIFADDR: |
6b96018b AB |
3447 | case SIOCGIFBRDADDR: |
3448 | case SIOCSIFBRDADDR: | |
3449 | case SIOCGIFDSTADDR: | |
3450 | case SIOCSIFDSTADDR: | |
3451 | case SIOCGIFNETMASK: | |
3452 | case SIOCSIFNETMASK: | |
3453 | case SIOCSIFPFLAGS: | |
3454 | case SIOCGIFPFLAGS: | |
3455 | case SIOCGIFTXQLEN: | |
3456 | case SIOCSIFTXQLEN: | |
3457 | case SIOCBRADDIF: | |
3458 | case SIOCBRDELIF: | |
c6c9fee3 | 3459 | case SIOCGIFNAME: |
9177efd3 AB |
3460 | case SIOCSIFNAME: |
3461 | case SIOCGMIIPHY: | |
3462 | case SIOCGMIIREG: | |
3463 | case SIOCSMIIREG: | |
f92d4fc9 AV |
3464 | case SIOCBONDENSLAVE: |
3465 | case SIOCBONDRELEASE: | |
3466 | case SIOCBONDSETHWADDR: | |
3467 | case SIOCBONDCHANGEACTIVE: | |
37ac39bd JB |
3468 | return compat_ifreq_ioctl(net, sock, cmd, argp); |
3469 | ||
6b96018b AB |
3470 | case SIOCSARP: |
3471 | case SIOCGARP: | |
3472 | case SIOCDARP: | |
6b96018b | 3473 | case SIOCATMARK: |
63ff03ab | 3474 | return sock_do_ioctl(net, sock, cmd, arg); |
9177efd3 AB |
3475 | } |
3476 | ||
6b96018b AB |
3477 | return -ENOIOCTLCMD; |
3478 | } | |
7a229387 | 3479 | |
95c96174 | 3480 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3481 | unsigned long arg) |
89bbfc95 SP |
3482 | { |
3483 | struct socket *sock = file->private_data; | |
3484 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3485 | struct sock *sk; |
3486 | struct net *net; | |
3487 | ||
3488 | sk = sock->sk; | |
3489 | net = sock_net(sk); | |
89bbfc95 SP |
3490 | |
3491 | if (sock->ops->compat_ioctl) | |
3492 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3493 | ||
87de87d5 DM |
3494 | if (ret == -ENOIOCTLCMD && |
3495 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3496 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3497 | ||
6b96018b AB |
3498 | if (ret == -ENOIOCTLCMD) |
3499 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3500 | ||
89bbfc95 SP |
3501 | return ret; |
3502 | } | |
3503 | #endif | |
3504 | ||
8a3c245c PT |
3505 | /** |
3506 | * kernel_bind - bind an address to a socket (kernel space) | |
3507 | * @sock: socket | |
3508 | * @addr: address | |
3509 | * @addrlen: length of address | |
3510 | * | |
3511 | * Returns 0 or an error. | |
3512 | */ | |
3513 | ||
ac5a488e SS |
3514 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3515 | { | |
3516 | return sock->ops->bind(sock, addr, addrlen); | |
3517 | } | |
c6d409cf | 3518 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e | 3519 | |
8a3c245c PT |
3520 | /** |
3521 | * kernel_listen - move socket to listening state (kernel space) | |
3522 | * @sock: socket | |
3523 | * @backlog: pending connections queue size | |
3524 | * | |
3525 | * Returns 0 or an error. | |
3526 | */ | |
3527 | ||
ac5a488e SS |
3528 | int kernel_listen(struct socket *sock, int backlog) |
3529 | { | |
3530 | return sock->ops->listen(sock, backlog); | |
3531 | } | |
c6d409cf | 3532 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e | 3533 | |
8a3c245c PT |
3534 | /** |
3535 | * kernel_accept - accept a connection (kernel space) | |
3536 | * @sock: listening socket | |
3537 | * @newsock: new connected socket | |
3538 | * @flags: flags | |
3539 | * | |
3540 | * @flags must be SOCK_CLOEXEC, SOCK_NONBLOCK or 0. | |
3541 | * If it fails, @newsock is guaranteed to be %NULL. | |
3542 | * Returns 0 or an error. | |
3543 | */ | |
3544 | ||
ac5a488e SS |
3545 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) |
3546 | { | |
3547 | struct sock *sk = sock->sk; | |
3548 | int err; | |
3549 | ||
3550 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3551 | newsock); | |
3552 | if (err < 0) | |
3553 | goto done; | |
3554 | ||
cdfbabfb | 3555 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3556 | if (err < 0) { |
3557 | sock_release(*newsock); | |
fa8705b0 | 3558 | *newsock = NULL; |
ac5a488e SS |
3559 | goto done; |
3560 | } | |
3561 | ||
3562 | (*newsock)->ops = sock->ops; | |
1b08534e | 3563 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3564 | |
3565 | done: | |
3566 | return err; | |
3567 | } | |
c6d409cf | 3568 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e | 3569 | |
8a3c245c PT |
3570 | /** |
3571 | * kernel_connect - connect a socket (kernel space) | |
3572 | * @sock: socket | |
3573 | * @addr: address | |
3574 | * @addrlen: address length | |
3575 | * @flags: flags (O_NONBLOCK, ...) | |
3576 | * | |
3577 | * For datagram sockets, @addr is the addres to which datagrams are sent | |
3578 | * by default, and the only address from which datagrams are received. | |
3579 | * For stream sockets, attempts to connect to @addr. | |
3580 | * Returns 0 or an error code. | |
3581 | */ | |
3582 | ||
ac5a488e | 3583 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, |
4768fbcb | 3584 | int flags) |
ac5a488e SS |
3585 | { |
3586 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3587 | } | |
c6d409cf | 3588 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e | 3589 | |
8a3c245c PT |
3590 | /** |
3591 | * kernel_getsockname - get the address which the socket is bound (kernel space) | |
3592 | * @sock: socket | |
3593 | * @addr: address holder | |
3594 | * | |
3595 | * Fills the @addr pointer with the address which the socket is bound. | |
3596 | * Returns 0 or an error code. | |
3597 | */ | |
3598 | ||
9b2c45d4 | 3599 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3600 | { |
9b2c45d4 | 3601 | return sock->ops->getname(sock, addr, 0); |
ac5a488e | 3602 | } |
c6d409cf | 3603 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e | 3604 | |
8a3c245c PT |
3605 | /** |
3606 | * kernel_peername - get the address which the socket is connected (kernel space) | |
3607 | * @sock: socket | |
3608 | * @addr: address holder | |
3609 | * | |
3610 | * Fills the @addr pointer with the address which the socket is connected. | |
3611 | * Returns 0 or an error code. | |
3612 | */ | |
3613 | ||
9b2c45d4 | 3614 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3615 | { |
9b2c45d4 | 3616 | return sock->ops->getname(sock, addr, 1); |
ac5a488e | 3617 | } |
c6d409cf | 3618 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e | 3619 | |
8a3c245c PT |
3620 | /** |
3621 | * kernel_getsockopt - get a socket option (kernel space) | |
3622 | * @sock: socket | |
3623 | * @level: API level (SOL_SOCKET, ...) | |
3624 | * @optname: option tag | |
3625 | * @optval: option value | |
3626 | * @optlen: option length | |
3627 | * | |
3628 | * Assigns the option length to @optlen. | |
3629 | * Returns 0 or an error. | |
3630 | */ | |
3631 | ||
ac5a488e SS |
3632 | int kernel_getsockopt(struct socket *sock, int level, int optname, |
3633 | char *optval, int *optlen) | |
3634 | { | |
3635 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3636 | char __user *uoptval; |
3637 | int __user *uoptlen; | |
ac5a488e SS |
3638 | int err; |
3639 | ||
fb8621bb NK |
3640 | uoptval = (char __user __force *) optval; |
3641 | uoptlen = (int __user __force *) optlen; | |
3642 | ||
ac5a488e SS |
3643 | set_fs(KERNEL_DS); |
3644 | if (level == SOL_SOCKET) | |
fb8621bb | 3645 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3646 | else |
fb8621bb NK |
3647 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3648 | uoptlen); | |
ac5a488e SS |
3649 | set_fs(oldfs); |
3650 | return err; | |
3651 | } | |
c6d409cf | 3652 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e | 3653 | |
8a3c245c PT |
3654 | /** |
3655 | * kernel_setsockopt - set a socket option (kernel space) | |
3656 | * @sock: socket | |
3657 | * @level: API level (SOL_SOCKET, ...) | |
3658 | * @optname: option tag | |
3659 | * @optval: option value | |
3660 | * @optlen: option length | |
3661 | * | |
3662 | * Returns 0 or an error. | |
3663 | */ | |
3664 | ||
ac5a488e | 3665 | int kernel_setsockopt(struct socket *sock, int level, int optname, |
b7058842 | 3666 | char *optval, unsigned int optlen) |
ac5a488e SS |
3667 | { |
3668 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3669 | char __user *uoptval; |
ac5a488e SS |
3670 | int err; |
3671 | ||
fb8621bb NK |
3672 | uoptval = (char __user __force *) optval; |
3673 | ||
ac5a488e SS |
3674 | set_fs(KERNEL_DS); |
3675 | if (level == SOL_SOCKET) | |
fb8621bb | 3676 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3677 | else |
fb8621bb | 3678 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3679 | optlen); |
3680 | set_fs(oldfs); | |
3681 | return err; | |
3682 | } | |
c6d409cf | 3683 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e | 3684 | |
8a3c245c PT |
3685 | /** |
3686 | * kernel_sendpage - send a &page through a socket (kernel space) | |
3687 | * @sock: socket | |
3688 | * @page: page | |
3689 | * @offset: page offset | |
3690 | * @size: total size in bytes | |
3691 | * @flags: flags (MSG_DONTWAIT, ...) | |
3692 | * | |
3693 | * Returns the total amount sent in bytes or an error. | |
3694 | */ | |
3695 | ||
ac5a488e SS |
3696 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, |
3697 | size_t size, int flags) | |
3698 | { | |
3699 | if (sock->ops->sendpage) | |
3700 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3701 | ||
3702 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3703 | } | |
c6d409cf | 3704 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3705 | |
8a3c245c PT |
3706 | /** |
3707 | * kernel_sendpage_locked - send a &page through the locked sock (kernel space) | |
3708 | * @sk: sock | |
3709 | * @page: page | |
3710 | * @offset: page offset | |
3711 | * @size: total size in bytes | |
3712 | * @flags: flags (MSG_DONTWAIT, ...) | |
3713 | * | |
3714 | * Returns the total amount sent in bytes or an error. | |
3715 | * Caller must hold @sk. | |
3716 | */ | |
3717 | ||
306b13eb TH |
3718 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3719 | size_t size, int flags) | |
3720 | { | |
3721 | struct socket *sock = sk->sk_socket; | |
3722 | ||
3723 | if (sock->ops->sendpage_locked) | |
3724 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3725 | flags); | |
3726 | ||
3727 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3728 | } | |
3729 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3730 | ||
8a3c245c PT |
3731 | /** |
3732 | * kernel_shutdown - shut down part of a full-duplex connection (kernel space) | |
3733 | * @sock: socket | |
3734 | * @how: connection part | |
3735 | * | |
3736 | * Returns 0 or an error. | |
3737 | */ | |
3738 | ||
91cf45f0 TM |
3739 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3740 | { | |
3741 | return sock->ops->shutdown(sock, how); | |
3742 | } | |
91cf45f0 | 3743 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 | 3744 | |
8a3c245c PT |
3745 | /** |
3746 | * kernel_sock_ip_overhead - returns the IP overhead imposed by a socket | |
3747 | * @sk: socket | |
3748 | * | |
3749 | * This routine returns the IP overhead imposed by a socket i.e. | |
3750 | * the length of the underlying IP header, depending on whether | |
3751 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
3752 | * on at the socket. Assumes that the caller has a lock on the socket. | |
113c3075 | 3753 | */ |
8a3c245c | 3754 | |
113c3075 P |
3755 | u32 kernel_sock_ip_overhead(struct sock *sk) |
3756 | { | |
3757 | struct inet_sock *inet; | |
3758 | struct ip_options_rcu *opt; | |
3759 | u32 overhead = 0; | |
113c3075 P |
3760 | #if IS_ENABLED(CONFIG_IPV6) |
3761 | struct ipv6_pinfo *np; | |
3762 | struct ipv6_txoptions *optv6 = NULL; | |
3763 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3764 | ||
3765 | if (!sk) | |
3766 | return overhead; | |
3767 | ||
113c3075 P |
3768 | switch (sk->sk_family) { |
3769 | case AF_INET: | |
3770 | inet = inet_sk(sk); | |
3771 | overhead += sizeof(struct iphdr); | |
3772 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3773 | sock_owned_by_user(sk)); |
113c3075 P |
3774 | if (opt) |
3775 | overhead += opt->opt.optlen; | |
3776 | return overhead; | |
3777 | #if IS_ENABLED(CONFIG_IPV6) | |
3778 | case AF_INET6: | |
3779 | np = inet6_sk(sk); | |
3780 | overhead += sizeof(struct ipv6hdr); | |
3781 | if (np) | |
3782 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3783 | sock_owned_by_user(sk)); |
113c3075 P |
3784 | if (optv6) |
3785 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3786 | return overhead; | |
3787 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3788 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3789 | return overhead; | |
3790 | } | |
3791 | } | |
3792 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |