]>
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; |
28407630 | 407 | return file; |
39d8c1b6 | 408 | } |
56b31d1c | 409 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 410 | |
56b31d1c | 411 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
412 | { |
413 | struct file *newfile; | |
28407630 | 414 | int fd = get_unused_fd_flags(flags); |
ce4bb04c AV |
415 | if (unlikely(fd < 0)) { |
416 | sock_release(sock); | |
28407630 | 417 | return fd; |
ce4bb04c | 418 | } |
39d8c1b6 | 419 | |
aab174f0 | 420 | newfile = sock_alloc_file(sock, flags, NULL); |
4546e44c | 421 | if (!IS_ERR(newfile)) { |
39d8c1b6 | 422 | fd_install(fd, newfile); |
28407630 AV |
423 | return fd; |
424 | } | |
7cbe66b6 | 425 | |
28407630 AV |
426 | put_unused_fd(fd); |
427 | return PTR_ERR(newfile); | |
1da177e4 LT |
428 | } |
429 | ||
8a3c245c PT |
430 | /** |
431 | * sock_from_file - Return the &socket bounded to @file. | |
432 | * @file: file | |
433 | * @err: pointer to an error code return | |
434 | * | |
435 | * On failure returns %NULL and assigns -ENOTSOCK to @err. | |
436 | */ | |
437 | ||
406a3c63 | 438 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 439 | { |
6cb153ca BL |
440 | if (file->f_op == &socket_file_ops) |
441 | return file->private_data; /* set in sock_map_fd */ | |
442 | ||
23bb80d2 ED |
443 | *err = -ENOTSOCK; |
444 | return NULL; | |
6cb153ca | 445 | } |
406a3c63 | 446 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 447 | |
1da177e4 | 448 | /** |
c6d409cf | 449 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
450 | * @fd: file handle |
451 | * @err: pointer to an error code return | |
452 | * | |
453 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 454 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
455 | * with a negative errno code and NULL is returned. The function checks |
456 | * for both invalid handles and passing a handle which is not a socket. | |
457 | * | |
458 | * On a success the socket object pointer is returned. | |
459 | */ | |
460 | ||
461 | struct socket *sockfd_lookup(int fd, int *err) | |
462 | { | |
463 | struct file *file; | |
1da177e4 LT |
464 | struct socket *sock; |
465 | ||
89bddce5 SH |
466 | file = fget(fd); |
467 | if (!file) { | |
1da177e4 LT |
468 | *err = -EBADF; |
469 | return NULL; | |
470 | } | |
89bddce5 | 471 | |
6cb153ca BL |
472 | sock = sock_from_file(file, err); |
473 | if (!sock) | |
1da177e4 | 474 | fput(file); |
6cb153ca BL |
475 | return sock; |
476 | } | |
c6d409cf | 477 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 478 | |
6cb153ca BL |
479 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
480 | { | |
00e188ef | 481 | struct fd f = fdget(fd); |
6cb153ca BL |
482 | struct socket *sock; |
483 | ||
3672558c | 484 | *err = -EBADF; |
00e188ef AV |
485 | if (f.file) { |
486 | sock = sock_from_file(f.file, err); | |
487 | if (likely(sock)) { | |
488 | *fput_needed = f.flags; | |
6cb153ca | 489 | return sock; |
00e188ef AV |
490 | } |
491 | fdput(f); | |
1da177e4 | 492 | } |
6cb153ca | 493 | return NULL; |
1da177e4 LT |
494 | } |
495 | ||
600e1779 MY |
496 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
497 | size_t size) | |
498 | { | |
499 | ssize_t len; | |
500 | ssize_t used = 0; | |
501 | ||
c5ef6035 | 502 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
503 | if (len < 0) |
504 | return len; | |
505 | used += len; | |
506 | if (buffer) { | |
507 | if (size < used) | |
508 | return -ERANGE; | |
509 | buffer += len; | |
510 | } | |
511 | ||
512 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
513 | used += len; | |
514 | if (buffer) { | |
515 | if (size < used) | |
516 | return -ERANGE; | |
517 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
518 | buffer += len; | |
519 | } | |
520 | ||
521 | return used; | |
522 | } | |
523 | ||
dc647ec8 | 524 | static int sockfs_setattr(struct dentry *dentry, struct iattr *iattr) |
86741ec2 LC |
525 | { |
526 | int err = simple_setattr(dentry, iattr); | |
527 | ||
e1a3a60a | 528 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
529 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
530 | ||
6d8c50dc CW |
531 | if (sock->sk) |
532 | sock->sk->sk_uid = iattr->ia_uid; | |
533 | else | |
534 | err = -ENOENT; | |
86741ec2 LC |
535 | } |
536 | ||
537 | return err; | |
538 | } | |
539 | ||
600e1779 | 540 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 541 | .listxattr = sockfs_listxattr, |
86741ec2 | 542 | .setattr = sockfs_setattr, |
600e1779 MY |
543 | }; |
544 | ||
1da177e4 | 545 | /** |
8a3c245c | 546 | * sock_alloc - allocate a socket |
89bddce5 | 547 | * |
1da177e4 LT |
548 | * Allocate a new inode and socket object. The two are bound together |
549 | * and initialised. The socket is then returned. If we are out of inodes | |
8a3c245c | 550 | * NULL is returned. This functions uses GFP_KERNEL internally. |
1da177e4 LT |
551 | */ |
552 | ||
f4a00aac | 553 | struct socket *sock_alloc(void) |
1da177e4 | 554 | { |
89bddce5 SH |
555 | struct inode *inode; |
556 | struct socket *sock; | |
1da177e4 | 557 | |
a209dfc7 | 558 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
559 | if (!inode) |
560 | return NULL; | |
561 | ||
562 | sock = SOCKET_I(inode); | |
563 | ||
85fe4025 | 564 | inode->i_ino = get_next_ino(); |
89bddce5 | 565 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
566 | inode->i_uid = current_fsuid(); |
567 | inode->i_gid = current_fsgid(); | |
600e1779 | 568 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 569 | |
1da177e4 LT |
570 | return sock; |
571 | } | |
f4a00aac | 572 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 573 | |
1da177e4 | 574 | /** |
8a3c245c | 575 | * sock_release - close a socket |
1da177e4 LT |
576 | * @sock: socket to close |
577 | * | |
578 | * The socket is released from the protocol stack if it has a release | |
579 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 580 | * an inode not a file. |
1da177e4 | 581 | */ |
89bddce5 | 582 | |
6d8c50dc | 583 | static void __sock_release(struct socket *sock, struct inode *inode) |
1da177e4 LT |
584 | { |
585 | if (sock->ops) { | |
586 | struct module *owner = sock->ops->owner; | |
587 | ||
6d8c50dc CW |
588 | if (inode) |
589 | inode_lock(inode); | |
1da177e4 | 590 | sock->ops->release(sock); |
ff7b11aa | 591 | sock->sk = NULL; |
6d8c50dc CW |
592 | if (inode) |
593 | inode_unlock(inode); | |
1da177e4 LT |
594 | sock->ops = NULL; |
595 | module_put(owner); | |
596 | } | |
597 | ||
333f7909 | 598 | if (sock->wq.fasync_list) |
3410f22e | 599 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 600 | |
1da177e4 LT |
601 | if (!sock->file) { |
602 | iput(SOCK_INODE(sock)); | |
603 | return; | |
604 | } | |
89bddce5 | 605 | sock->file = NULL; |
1da177e4 | 606 | } |
6d8c50dc CW |
607 | |
608 | void sock_release(struct socket *sock) | |
609 | { | |
610 | __sock_release(sock, NULL); | |
611 | } | |
c6d409cf | 612 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 613 | |
c14ac945 | 614 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 615 | { |
140c55d4 ED |
616 | u8 flags = *tx_flags; |
617 | ||
c14ac945 | 618 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
619 | flags |= SKBTX_HW_TSTAMP; |
620 | ||
c14ac945 | 621 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
622 | flags |= SKBTX_SW_TSTAMP; |
623 | ||
c14ac945 | 624 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
625 | flags |= SKBTX_SCHED_TSTAMP; |
626 | ||
140c55d4 | 627 | *tx_flags = flags; |
20d49473 | 628 | } |
67cc0d40 | 629 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 630 | |
8c3c447b PA |
631 | INDIRECT_CALLABLE_DECLARE(int inet_sendmsg(struct socket *, struct msghdr *, |
632 | size_t)); | |
a648a592 PA |
633 | INDIRECT_CALLABLE_DECLARE(int inet6_sendmsg(struct socket *, struct msghdr *, |
634 | size_t)); | |
d8725c86 | 635 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 636 | { |
a648a592 PA |
637 | int ret = INDIRECT_CALL_INET(sock->ops->sendmsg, inet6_sendmsg, |
638 | inet_sendmsg, sock, msg, | |
639 | msg_data_left(msg)); | |
d8725c86 AV |
640 | BUG_ON(ret == -EIOCBQUEUED); |
641 | return ret; | |
1da177e4 LT |
642 | } |
643 | ||
85806af0 RD |
644 | /** |
645 | * sock_sendmsg - send a message through @sock | |
646 | * @sock: socket | |
647 | * @msg: message to send | |
648 | * | |
649 | * Sends @msg through @sock, passing through LSM. | |
650 | * Returns the number of bytes sent, or an error code. | |
651 | */ | |
d8725c86 | 652 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 653 | { |
d8725c86 | 654 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 655 | msg_data_left(msg)); |
228e548e | 656 | |
d8725c86 | 657 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 658 | } |
c6d409cf | 659 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 | 660 | |
8a3c245c PT |
661 | /** |
662 | * kernel_sendmsg - send a message through @sock (kernel-space) | |
663 | * @sock: socket | |
664 | * @msg: message header | |
665 | * @vec: kernel vec | |
666 | * @num: vec array length | |
667 | * @size: total message data size | |
668 | * | |
669 | * Builds the message data with @vec and sends it through @sock. | |
670 | * Returns the number of bytes sent, or an error code. | |
671 | */ | |
672 | ||
1da177e4 LT |
673 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
674 | struct kvec *vec, size_t num, size_t size) | |
675 | { | |
aa563d7b | 676 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
d8725c86 | 677 | return sock_sendmsg(sock, msg); |
1da177e4 | 678 | } |
c6d409cf | 679 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 680 | |
8a3c245c PT |
681 | /** |
682 | * kernel_sendmsg_locked - send a message through @sock (kernel-space) | |
683 | * @sk: sock | |
684 | * @msg: message header | |
685 | * @vec: output s/g array | |
686 | * @num: output s/g array length | |
687 | * @size: total message data size | |
688 | * | |
689 | * Builds the message data with @vec and sends it through @sock. | |
690 | * Returns the number of bytes sent, or an error code. | |
691 | * Caller must hold @sk. | |
692 | */ | |
693 | ||
306b13eb TH |
694 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
695 | struct kvec *vec, size_t num, size_t size) | |
696 | { | |
697 | struct socket *sock = sk->sk_socket; | |
698 | ||
699 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 700 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb | 701 | |
aa563d7b | 702 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
306b13eb TH |
703 | |
704 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
705 | } | |
706 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
707 | ||
8605330a SHY |
708 | static bool skb_is_err_queue(const struct sk_buff *skb) |
709 | { | |
710 | /* pkt_type of skbs enqueued on the error queue are set to | |
711 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
712 | * in recvmsg, since skbs received on a local socket will never | |
713 | * have a pkt_type of PACKET_OUTGOING. | |
714 | */ | |
715 | return skb->pkt_type == PACKET_OUTGOING; | |
716 | } | |
717 | ||
b50a5c70 ML |
718 | /* On transmit, software and hardware timestamps are returned independently. |
719 | * As the two skb clones share the hardware timestamp, which may be updated | |
720 | * before the software timestamp is received, a hardware TX timestamp may be | |
721 | * returned only if there is no software TX timestamp. Ignore false software | |
722 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
7f1bc6e9 | 723 | * option SO_TIMESTAMP_OLD(NS) is enabled on the socket, even when the skb has a |
b50a5c70 ML |
724 | * hardware timestamp. |
725 | */ | |
726 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
727 | { | |
728 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
729 | } | |
730 | ||
aad9c8c4 ML |
731 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb) |
732 | { | |
733 | struct scm_ts_pktinfo ts_pktinfo; | |
734 | struct net_device *orig_dev; | |
735 | ||
736 | if (!skb_mac_header_was_set(skb)) | |
737 | return; | |
738 | ||
739 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
740 | ||
741 | rcu_read_lock(); | |
742 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
743 | if (orig_dev) | |
744 | ts_pktinfo.if_index = orig_dev->ifindex; | |
745 | rcu_read_unlock(); | |
746 | ||
747 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
748 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
749 | sizeof(ts_pktinfo), &ts_pktinfo); | |
750 | } | |
751 | ||
92f37fd2 ED |
752 | /* |
753 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
754 | */ | |
755 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
756 | struct sk_buff *skb) | |
757 | { | |
20d49473 | 758 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
887feae3 | 759 | int new_tstamp = sock_flag(sk, SOCK_TSTAMP_NEW); |
9718475e DD |
760 | struct scm_timestamping_internal tss; |
761 | ||
b50a5c70 | 762 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
763 | struct skb_shared_hwtstamps *shhwtstamps = |
764 | skb_hwtstamps(skb); | |
765 | ||
766 | /* Race occurred between timestamp enabling and packet | |
767 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 768 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 769 | __net_timestamp(skb); |
b50a5c70 ML |
770 | false_tstamp = 1; |
771 | } | |
20d49473 PO |
772 | |
773 | if (need_software_tstamp) { | |
774 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
887feae3 DD |
775 | if (new_tstamp) { |
776 | struct __kernel_sock_timeval tv; | |
777 | ||
778 | skb_get_new_timestamp(skb, &tv); | |
779 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_NEW, | |
780 | sizeof(tv), &tv); | |
781 | } else { | |
782 | struct __kernel_old_timeval tv; | |
783 | ||
784 | skb_get_timestamp(skb, &tv); | |
785 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_OLD, | |
786 | sizeof(tv), &tv); | |
787 | } | |
20d49473 | 788 | } else { |
887feae3 DD |
789 | if (new_tstamp) { |
790 | struct __kernel_timespec ts; | |
791 | ||
792 | skb_get_new_timestampns(skb, &ts); | |
793 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_NEW, | |
794 | sizeof(ts), &ts); | |
795 | } else { | |
796 | struct timespec ts; | |
797 | ||
798 | skb_get_timestampns(skb, &ts); | |
799 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_OLD, | |
800 | sizeof(ts), &ts); | |
801 | } | |
20d49473 PO |
802 | } |
803 | } | |
804 | ||
f24b9be5 | 805 | memset(&tss, 0, sizeof(tss)); |
c199105d | 806 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
9718475e | 807 | ktime_to_timespec64_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 808 | empty = 0; |
4d276eb6 | 809 | if (shhwtstamps && |
b9f40e21 | 810 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
b50a5c70 | 811 | !skb_is_swtx_tstamp(skb, false_tstamp) && |
9718475e | 812 | ktime_to_timespec64_cond(shhwtstamps->hwtstamp, tss.ts + 2)) { |
4d276eb6 | 813 | empty = 0; |
aad9c8c4 ML |
814 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && |
815 | !skb_is_err_queue(skb)) | |
816 | put_ts_pktinfo(msg, skb); | |
817 | } | |
1c885808 | 818 | if (!empty) { |
9718475e DD |
819 | if (sock_flag(sk, SOCK_TSTAMP_NEW)) |
820 | put_cmsg_scm_timestamping64(msg, &tss); | |
821 | else | |
822 | put_cmsg_scm_timestamping(msg, &tss); | |
1c885808 | 823 | |
8605330a | 824 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 825 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
826 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
827 | skb->len, skb->data); | |
828 | } | |
92f37fd2 | 829 | } |
7c81fd8b ACM |
830 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
831 | ||
6e3e939f JB |
832 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
833 | struct sk_buff *skb) | |
834 | { | |
835 | int ack; | |
836 | ||
837 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
838 | return; | |
839 | if (!skb->wifi_acked_valid) | |
840 | return; | |
841 | ||
842 | ack = skb->wifi_acked; | |
843 | ||
844 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
845 | } | |
846 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
847 | ||
11165f14 | 848 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
849 | struct sk_buff *skb) | |
3b885787 | 850 | { |
744d5a3e | 851 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 852 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 853 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
854 | } |
855 | ||
767dd033 | 856 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
857 | struct sk_buff *skb) |
858 | { | |
859 | sock_recv_timestamp(msg, sk, skb); | |
860 | sock_recv_drops(msg, sk, skb); | |
861 | } | |
767dd033 | 862 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 863 | |
8c3c447b | 864 | INDIRECT_CALLABLE_DECLARE(int inet_recvmsg(struct socket *, struct msghdr *, |
a648a592 PA |
865 | size_t, int)); |
866 | INDIRECT_CALLABLE_DECLARE(int inet6_recvmsg(struct socket *, struct msghdr *, | |
867 | size_t, int)); | |
1b784140 | 868 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 869 | int flags) |
1da177e4 | 870 | { |
a648a592 PA |
871 | return INDIRECT_CALL_INET(sock->ops->recvmsg, inet6_recvmsg, |
872 | inet_recvmsg, sock, msg, msg_data_left(msg), | |
873 | flags); | |
1da177e4 LT |
874 | } |
875 | ||
85806af0 RD |
876 | /** |
877 | * sock_recvmsg - receive a message from @sock | |
878 | * @sock: socket | |
879 | * @msg: message to receive | |
880 | * @flags: message flags | |
881 | * | |
882 | * Receives @msg from @sock, passing through LSM. Returns the total number | |
883 | * of bytes received, or an error. | |
884 | */ | |
2da62906 | 885 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 886 | { |
2da62906 | 887 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 888 | |
2da62906 | 889 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 890 | } |
c6d409cf | 891 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 892 | |
c1249c0a | 893 | /** |
8a3c245c PT |
894 | * kernel_recvmsg - Receive a message from a socket (kernel space) |
895 | * @sock: The socket to receive the message from | |
896 | * @msg: Received message | |
897 | * @vec: Input s/g array for message data | |
898 | * @num: Size of input s/g array | |
899 | * @size: Number of bytes to read | |
900 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
c1249c0a | 901 | * |
8a3c245c PT |
902 | * On return the msg structure contains the scatter/gather array passed in the |
903 | * vec argument. The array is modified so that it consists of the unfilled | |
904 | * portion of the original array. | |
c1249c0a | 905 | * |
8a3c245c | 906 | * The returned value is the total number of bytes received, or an error. |
c1249c0a | 907 | */ |
8a3c245c | 908 | |
89bddce5 SH |
909 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
910 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
911 | { |
912 | mm_segment_t oldfs = get_fs(); | |
913 | int result; | |
914 | ||
aa563d7b | 915 | iov_iter_kvec(&msg->msg_iter, READ, vec, num, size); |
1da177e4 | 916 | set_fs(KERNEL_DS); |
2da62906 | 917 | result = sock_recvmsg(sock, msg, flags); |
1da177e4 LT |
918 | set_fs(oldfs); |
919 | return result; | |
920 | } | |
c6d409cf | 921 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 922 | |
ce1d4d3e CH |
923 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
924 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 925 | { |
1da177e4 LT |
926 | struct socket *sock; |
927 | int flags; | |
928 | ||
ce1d4d3e CH |
929 | sock = file->private_data; |
930 | ||
35f9c09f ED |
931 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
932 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
933 | flags |= more; | |
ce1d4d3e | 934 | |
e6949583 | 935 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 936 | } |
1da177e4 | 937 | |
9c55e01c | 938 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 939 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
940 | unsigned int flags) |
941 | { | |
942 | struct socket *sock = file->private_data; | |
943 | ||
997b37da | 944 | if (unlikely(!sock->ops->splice_read)) |
95506588 | 945 | return generic_file_splice_read(file, ppos, pipe, len, flags); |
997b37da | 946 | |
9c55e01c JA |
947 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
948 | } | |
949 | ||
8ae5e030 | 950 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 951 | { |
6d652330 AV |
952 | struct file *file = iocb->ki_filp; |
953 | struct socket *sock = file->private_data; | |
0345f931 | 954 | struct msghdr msg = {.msg_iter = *to, |
955 | .msg_iocb = iocb}; | |
8ae5e030 | 956 | ssize_t res; |
ce1d4d3e | 957 | |
8ae5e030 AV |
958 | if (file->f_flags & O_NONBLOCK) |
959 | msg.msg_flags = MSG_DONTWAIT; | |
960 | ||
961 | if (iocb->ki_pos != 0) | |
1da177e4 | 962 | return -ESPIPE; |
027445c3 | 963 | |
66ee59af | 964 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
965 | return 0; |
966 | ||
2da62906 | 967 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
968 | *to = msg.msg_iter; |
969 | return res; | |
1da177e4 LT |
970 | } |
971 | ||
8ae5e030 | 972 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 973 | { |
6d652330 AV |
974 | struct file *file = iocb->ki_filp; |
975 | struct socket *sock = file->private_data; | |
0345f931 | 976 | struct msghdr msg = {.msg_iter = *from, |
977 | .msg_iocb = iocb}; | |
8ae5e030 | 978 | ssize_t res; |
1da177e4 | 979 | |
8ae5e030 | 980 | if (iocb->ki_pos != 0) |
ce1d4d3e | 981 | return -ESPIPE; |
027445c3 | 982 | |
8ae5e030 AV |
983 | if (file->f_flags & O_NONBLOCK) |
984 | msg.msg_flags = MSG_DONTWAIT; | |
985 | ||
6d652330 AV |
986 | if (sock->type == SOCK_SEQPACKET) |
987 | msg.msg_flags |= MSG_EOR; | |
988 | ||
d8725c86 | 989 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
990 | *from = msg.msg_iter; |
991 | return res; | |
1da177e4 LT |
992 | } |
993 | ||
1da177e4 LT |
994 | /* |
995 | * Atomic setting of ioctl hooks to avoid race | |
996 | * with module unload. | |
997 | */ | |
998 | ||
4a3e2f71 | 999 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 1000 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 1001 | |
881d966b | 1002 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 1003 | { |
4a3e2f71 | 1004 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 1005 | br_ioctl_hook = hook; |
4a3e2f71 | 1006 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1007 | } |
1008 | EXPORT_SYMBOL(brioctl_set); | |
1009 | ||
4a3e2f71 | 1010 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 1011 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 1012 | |
881d966b | 1013 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 1014 | { |
4a3e2f71 | 1015 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1016 | vlan_ioctl_hook = hook; |
4a3e2f71 | 1017 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
1018 | } |
1019 | EXPORT_SYMBOL(vlan_ioctl_set); | |
1020 | ||
4a3e2f71 | 1021 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 1022 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 1023 | |
89bddce5 | 1024 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 1025 | { |
4a3e2f71 | 1026 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 1027 | dlci_ioctl_hook = hook; |
4a3e2f71 | 1028 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
1029 | } |
1030 | EXPORT_SYMBOL(dlci_ioctl_set); | |
1031 | ||
6b96018b | 1032 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
63ff03ab | 1033 | unsigned int cmd, unsigned long arg) |
6b96018b AB |
1034 | { |
1035 | int err; | |
1036 | void __user *argp = (void __user *)arg; | |
1037 | ||
1038 | err = sock->ops->ioctl(sock, cmd, arg); | |
1039 | ||
1040 | /* | |
1041 | * If this ioctl is unknown try to hand it down | |
1042 | * to the NIC driver. | |
1043 | */ | |
36fd633e AV |
1044 | if (err != -ENOIOCTLCMD) |
1045 | return err; | |
6b96018b | 1046 | |
36fd633e AV |
1047 | if (cmd == SIOCGIFCONF) { |
1048 | struct ifconf ifc; | |
1049 | if (copy_from_user(&ifc, argp, sizeof(struct ifconf))) | |
1050 | return -EFAULT; | |
1051 | rtnl_lock(); | |
1052 | err = dev_ifconf(net, &ifc, sizeof(struct ifreq)); | |
1053 | rtnl_unlock(); | |
1054 | if (!err && copy_to_user(argp, &ifc, sizeof(struct ifconf))) | |
1055 | err = -EFAULT; | |
44c02a2c AV |
1056 | } else { |
1057 | struct ifreq ifr; | |
1058 | bool need_copyout; | |
63ff03ab | 1059 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) |
44c02a2c AV |
1060 | return -EFAULT; |
1061 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1062 | if (!err && need_copyout) | |
63ff03ab | 1063 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) |
44c02a2c | 1064 | return -EFAULT; |
36fd633e | 1065 | } |
6b96018b AB |
1066 | return err; |
1067 | } | |
1068 | ||
1da177e4 LT |
1069 | /* |
1070 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1071 | * what to do with it - that's up to the protocol still. | |
1072 | */ | |
1073 | ||
8a3c245c PT |
1074 | /** |
1075 | * get_net_ns - increment the refcount of the network namespace | |
1076 | * @ns: common namespace (net) | |
1077 | * | |
1078 | * Returns the net's common namespace. | |
1079 | */ | |
1080 | ||
d8d211a2 | 1081 | struct ns_common *get_net_ns(struct ns_common *ns) |
c62cce2c AV |
1082 | { |
1083 | return &get_net(container_of(ns, struct net, ns))->ns; | |
1084 | } | |
d8d211a2 | 1085 | EXPORT_SYMBOL_GPL(get_net_ns); |
c62cce2c | 1086 | |
1da177e4 LT |
1087 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
1088 | { | |
1089 | struct socket *sock; | |
881d966b | 1090 | struct sock *sk; |
1da177e4 LT |
1091 | void __user *argp = (void __user *)arg; |
1092 | int pid, err; | |
881d966b | 1093 | struct net *net; |
1da177e4 | 1094 | |
b69aee04 | 1095 | sock = file->private_data; |
881d966b | 1096 | sk = sock->sk; |
3b1e0a65 | 1097 | net = sock_net(sk); |
44c02a2c AV |
1098 | if (unlikely(cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))) { |
1099 | struct ifreq ifr; | |
1100 | bool need_copyout; | |
1101 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
1102 | return -EFAULT; | |
1103 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1104 | if (!err && need_copyout) | |
1105 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
1106 | return -EFAULT; | |
1da177e4 | 1107 | } else |
3d23e349 | 1108 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1109 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
b1b0c245 | 1110 | err = wext_handle_ioctl(net, cmd, argp); |
1da177e4 | 1111 | } else |
3d23e349 | 1112 | #endif |
89bddce5 | 1113 | switch (cmd) { |
1da177e4 LT |
1114 | case FIOSETOWN: |
1115 | case SIOCSPGRP: | |
1116 | err = -EFAULT; | |
1117 | if (get_user(pid, (int __user *)argp)) | |
1118 | break; | |
393cc3f5 | 1119 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1120 | break; |
1121 | case FIOGETOWN: | |
1122 | case SIOCGPGRP: | |
609d7fa9 | 1123 | err = put_user(f_getown(sock->file), |
89bddce5 | 1124 | (int __user *)argp); |
1da177e4 LT |
1125 | break; |
1126 | case SIOCGIFBR: | |
1127 | case SIOCSIFBR: | |
1128 | case SIOCBRADDBR: | |
1129 | case SIOCBRDELBR: | |
1130 | err = -ENOPKG; | |
1131 | if (!br_ioctl_hook) | |
1132 | request_module("bridge"); | |
1133 | ||
4a3e2f71 | 1134 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1135 | if (br_ioctl_hook) |
881d966b | 1136 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1137 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1138 | break; |
1139 | case SIOCGIFVLAN: | |
1140 | case SIOCSIFVLAN: | |
1141 | err = -ENOPKG; | |
1142 | if (!vlan_ioctl_hook) | |
1143 | request_module("8021q"); | |
1144 | ||
4a3e2f71 | 1145 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1146 | if (vlan_ioctl_hook) |
881d966b | 1147 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1148 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1149 | break; |
1da177e4 LT |
1150 | case SIOCADDDLCI: |
1151 | case SIOCDELDLCI: | |
1152 | err = -ENOPKG; | |
1153 | if (!dlci_ioctl_hook) | |
1154 | request_module("dlci"); | |
1155 | ||
7512cbf6 PE |
1156 | mutex_lock(&dlci_ioctl_mutex); |
1157 | if (dlci_ioctl_hook) | |
1da177e4 | 1158 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1159 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 | 1160 | break; |
c62cce2c AV |
1161 | case SIOCGSKNS: |
1162 | err = -EPERM; | |
1163 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1164 | break; | |
1165 | ||
1166 | err = open_related_ns(&net->ns, get_net_ns); | |
1167 | break; | |
0768e170 AB |
1168 | case SIOCGSTAMP_OLD: |
1169 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
1170 | if (!sock->ops->gettstamp) { |
1171 | err = -ENOIOCTLCMD; | |
1172 | break; | |
1173 | } | |
1174 | err = sock->ops->gettstamp(sock, argp, | |
0768e170 AB |
1175 | cmd == SIOCGSTAMP_OLD, |
1176 | !IS_ENABLED(CONFIG_64BIT)); | |
60747828 | 1177 | break; |
0768e170 AB |
1178 | case SIOCGSTAMP_NEW: |
1179 | case SIOCGSTAMPNS_NEW: | |
1180 | if (!sock->ops->gettstamp) { | |
1181 | err = -ENOIOCTLCMD; | |
1182 | break; | |
1183 | } | |
1184 | err = sock->ops->gettstamp(sock, argp, | |
1185 | cmd == SIOCGSTAMP_NEW, | |
1186 | false); | |
c7cbdbf2 | 1187 | break; |
1da177e4 | 1188 | default: |
63ff03ab | 1189 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1190 | break; |
89bddce5 | 1191 | } |
1da177e4 LT |
1192 | return err; |
1193 | } | |
1194 | ||
8a3c245c PT |
1195 | /** |
1196 | * sock_create_lite - creates a socket | |
1197 | * @family: protocol family (AF_INET, ...) | |
1198 | * @type: communication type (SOCK_STREAM, ...) | |
1199 | * @protocol: protocol (0, ...) | |
1200 | * @res: new socket | |
1201 | * | |
1202 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1203 | * The new socket initialization is not complete, see kernel_accept(). | |
1204 | * Returns 0 or an error. On failure @res is set to %NULL. | |
1205 | * This function internally uses GFP_KERNEL. | |
1206 | */ | |
1207 | ||
1da177e4 LT |
1208 | int sock_create_lite(int family, int type, int protocol, struct socket **res) |
1209 | { | |
1210 | int err; | |
1211 | struct socket *sock = NULL; | |
89bddce5 | 1212 | |
1da177e4 LT |
1213 | err = security_socket_create(family, type, protocol, 1); |
1214 | if (err) | |
1215 | goto out; | |
1216 | ||
1217 | sock = sock_alloc(); | |
1218 | if (!sock) { | |
1219 | err = -ENOMEM; | |
1220 | goto out; | |
1221 | } | |
1222 | ||
1da177e4 | 1223 | sock->type = type; |
7420ed23 VY |
1224 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1225 | if (err) | |
1226 | goto out_release; | |
1227 | ||
1da177e4 LT |
1228 | out: |
1229 | *res = sock; | |
1230 | return err; | |
7420ed23 VY |
1231 | out_release: |
1232 | sock_release(sock); | |
1233 | sock = NULL; | |
1234 | goto out; | |
1da177e4 | 1235 | } |
c6d409cf | 1236 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1237 | |
1238 | /* No kernel lock held - perfect */ | |
ade994f4 | 1239 | static __poll_t sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1240 | { |
3cafb376 | 1241 | struct socket *sock = file->private_data; |
a331de3b | 1242 | __poll_t events = poll_requested_events(wait), flag = 0; |
2d48d67f | 1243 | |
e88958e6 CH |
1244 | if (!sock->ops->poll) |
1245 | return 0; | |
f641f13b | 1246 | |
a331de3b CH |
1247 | if (sk_can_busy_loop(sock->sk)) { |
1248 | /* poll once if requested by the syscall */ | |
1249 | if (events & POLL_BUSY_LOOP) | |
1250 | sk_busy_loop(sock->sk, 1); | |
1251 | ||
1252 | /* if this socket can poll_ll, tell the system call */ | |
1253 | flag = POLL_BUSY_LOOP; | |
1254 | } | |
1255 | ||
1256 | return sock->ops->poll(file, sock, wait) | flag; | |
1da177e4 LT |
1257 | } |
1258 | ||
89bddce5 | 1259 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1260 | { |
b69aee04 | 1261 | struct socket *sock = file->private_data; |
1da177e4 LT |
1262 | |
1263 | return sock->ops->mmap(file, sock, vma); | |
1264 | } | |
1265 | ||
20380731 | 1266 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1267 | { |
6d8c50dc | 1268 | __sock_release(SOCKET_I(inode), inode); |
1da177e4 LT |
1269 | return 0; |
1270 | } | |
1271 | ||
1272 | /* | |
1273 | * Update the socket async list | |
1274 | * | |
1275 | * Fasync_list locking strategy. | |
1276 | * | |
1277 | * 1. fasync_list is modified only under process context socket lock | |
1278 | * i.e. under semaphore. | |
1279 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1280 | * or under socket lock |
1da177e4 LT |
1281 | */ |
1282 | ||
1283 | static int sock_fasync(int fd, struct file *filp, int on) | |
1284 | { | |
989a2979 ED |
1285 | struct socket *sock = filp->private_data; |
1286 | struct sock *sk = sock->sk; | |
333f7909 | 1287 | struct socket_wq *wq = &sock->wq; |
1da177e4 | 1288 | |
989a2979 | 1289 | if (sk == NULL) |
1da177e4 | 1290 | return -EINVAL; |
1da177e4 LT |
1291 | |
1292 | lock_sock(sk); | |
eaefd110 | 1293 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1294 | |
eaefd110 | 1295 | if (!wq->fasync_list) |
989a2979 ED |
1296 | sock_reset_flag(sk, SOCK_FASYNC); |
1297 | else | |
bcdce719 | 1298 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1299 | |
989a2979 | 1300 | release_sock(sk); |
1da177e4 LT |
1301 | return 0; |
1302 | } | |
1303 | ||
ceb5d58b | 1304 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1305 | |
ceb5d58b | 1306 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1307 | { |
ceb5d58b | 1308 | if (!wq || !wq->fasync_list) |
1da177e4 | 1309 | return -1; |
ceb5d58b | 1310 | |
89bddce5 | 1311 | switch (how) { |
8d8ad9d7 | 1312 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1313 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1314 | break; |
1315 | goto call_kill; | |
8d8ad9d7 | 1316 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1317 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 LT |
1318 | break; |
1319 | /* fall through */ | |
8d8ad9d7 | 1320 | case SOCK_WAKE_IO: |
89bddce5 | 1321 | call_kill: |
43815482 | 1322 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1323 | break; |
8d8ad9d7 | 1324 | case SOCK_WAKE_URG: |
43815482 | 1325 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1326 | } |
ceb5d58b | 1327 | |
1da177e4 LT |
1328 | return 0; |
1329 | } | |
c6d409cf | 1330 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1331 | |
8a3c245c PT |
1332 | /** |
1333 | * __sock_create - creates a socket | |
1334 | * @net: net namespace | |
1335 | * @family: protocol family (AF_INET, ...) | |
1336 | * @type: communication type (SOCK_STREAM, ...) | |
1337 | * @protocol: protocol (0, ...) | |
1338 | * @res: new socket | |
1339 | * @kern: boolean for kernel space sockets | |
1340 | * | |
1341 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1342 | * Returns 0 or an error. On failure @res is set to %NULL. @kern must | |
1343 | * be set to true if the socket resides in kernel space. | |
1344 | * This function internally uses GFP_KERNEL. | |
1345 | */ | |
1346 | ||
721db93a | 1347 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1348 | struct socket **res, int kern) |
1da177e4 LT |
1349 | { |
1350 | int err; | |
1351 | struct socket *sock; | |
55737fda | 1352 | const struct net_proto_family *pf; |
1da177e4 LT |
1353 | |
1354 | /* | |
89bddce5 | 1355 | * Check protocol is in range |
1da177e4 LT |
1356 | */ |
1357 | if (family < 0 || family >= NPROTO) | |
1358 | return -EAFNOSUPPORT; | |
1359 | if (type < 0 || type >= SOCK_MAX) | |
1360 | return -EINVAL; | |
1361 | ||
1362 | /* Compatibility. | |
1363 | ||
1364 | This uglymoron is moved from INET layer to here to avoid | |
1365 | deadlock in module load. | |
1366 | */ | |
1367 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1368 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1369 | current->comm); | |
1da177e4 LT |
1370 | family = PF_PACKET; |
1371 | } | |
1372 | ||
1373 | err = security_socket_create(family, type, protocol, kern); | |
1374 | if (err) | |
1375 | return err; | |
89bddce5 | 1376 | |
55737fda SH |
1377 | /* |
1378 | * Allocate the socket and allow the family to set things up. if | |
1379 | * the protocol is 0, the family is instructed to select an appropriate | |
1380 | * default. | |
1381 | */ | |
1382 | sock = sock_alloc(); | |
1383 | if (!sock) { | |
e87cc472 | 1384 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1385 | return -ENFILE; /* Not exactly a match, but its the |
1386 | closest posix thing */ | |
1387 | } | |
1388 | ||
1389 | sock->type = type; | |
1390 | ||
95a5afca | 1391 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1392 | /* Attempt to load a protocol module if the find failed. |
1393 | * | |
1394 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1395 | * requested real, full-featured networking support upon configuration. |
1396 | * Otherwise module support will break! | |
1397 | */ | |
190683a9 | 1398 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1399 | request_module("net-pf-%d", family); |
1da177e4 LT |
1400 | #endif |
1401 | ||
55737fda SH |
1402 | rcu_read_lock(); |
1403 | pf = rcu_dereference(net_families[family]); | |
1404 | err = -EAFNOSUPPORT; | |
1405 | if (!pf) | |
1406 | goto out_release; | |
1da177e4 LT |
1407 | |
1408 | /* | |
1409 | * We will call the ->create function, that possibly is in a loadable | |
1410 | * module, so we have to bump that loadable module refcnt first. | |
1411 | */ | |
55737fda | 1412 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1413 | goto out_release; |
1414 | ||
55737fda SH |
1415 | /* Now protected by module ref count */ |
1416 | rcu_read_unlock(); | |
1417 | ||
3f378b68 | 1418 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1419 | if (err < 0) |
1da177e4 | 1420 | goto out_module_put; |
a79af59e | 1421 | |
1da177e4 LT |
1422 | /* |
1423 | * Now to bump the refcnt of the [loadable] module that owns this | |
1424 | * socket at sock_release time we decrement its refcnt. | |
1425 | */ | |
55737fda SH |
1426 | if (!try_module_get(sock->ops->owner)) |
1427 | goto out_module_busy; | |
1428 | ||
1da177e4 LT |
1429 | /* |
1430 | * Now that we're done with the ->create function, the [loadable] | |
1431 | * module can have its refcnt decremented | |
1432 | */ | |
55737fda | 1433 | module_put(pf->owner); |
7420ed23 VY |
1434 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1435 | if (err) | |
3b185525 | 1436 | goto out_sock_release; |
55737fda | 1437 | *res = sock; |
1da177e4 | 1438 | |
55737fda SH |
1439 | return 0; |
1440 | ||
1441 | out_module_busy: | |
1442 | err = -EAFNOSUPPORT; | |
1da177e4 | 1443 | out_module_put: |
55737fda SH |
1444 | sock->ops = NULL; |
1445 | module_put(pf->owner); | |
1446 | out_sock_release: | |
1da177e4 | 1447 | sock_release(sock); |
55737fda SH |
1448 | return err; |
1449 | ||
1450 | out_release: | |
1451 | rcu_read_unlock(); | |
1452 | goto out_sock_release; | |
1da177e4 | 1453 | } |
721db93a | 1454 | EXPORT_SYMBOL(__sock_create); |
1da177e4 | 1455 | |
8a3c245c PT |
1456 | /** |
1457 | * sock_create - creates a socket | |
1458 | * @family: protocol family (AF_INET, ...) | |
1459 | * @type: communication type (SOCK_STREAM, ...) | |
1460 | * @protocol: protocol (0, ...) | |
1461 | * @res: new socket | |
1462 | * | |
1463 | * A wrapper around __sock_create(). | |
1464 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1465 | */ | |
1466 | ||
1da177e4 LT |
1467 | int sock_create(int family, int type, int protocol, struct socket **res) |
1468 | { | |
1b8d7ae4 | 1469 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1470 | } |
c6d409cf | 1471 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1472 | |
8a3c245c PT |
1473 | /** |
1474 | * sock_create_kern - creates a socket (kernel space) | |
1475 | * @net: net namespace | |
1476 | * @family: protocol family (AF_INET, ...) | |
1477 | * @type: communication type (SOCK_STREAM, ...) | |
1478 | * @protocol: protocol (0, ...) | |
1479 | * @res: new socket | |
1480 | * | |
1481 | * A wrapper around __sock_create(). | |
1482 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1483 | */ | |
1484 | ||
eeb1bd5c | 1485 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1486 | { |
eeb1bd5c | 1487 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1488 | } |
c6d409cf | 1489 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1490 | |
9d6a15c3 | 1491 | int __sys_socket(int family, int type, int protocol) |
1da177e4 LT |
1492 | { |
1493 | int retval; | |
1494 | struct socket *sock; | |
a677a039 UD |
1495 | int flags; |
1496 | ||
e38b36f3 UD |
1497 | /* Check the SOCK_* constants for consistency. */ |
1498 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1499 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1500 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1501 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1502 | ||
a677a039 | 1503 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1504 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1505 | return -EINVAL; |
1506 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1507 | |
aaca0bdc UD |
1508 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1509 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1510 | ||
1da177e4 LT |
1511 | retval = sock_create(family, type, protocol, &sock); |
1512 | if (retval < 0) | |
8e1611e2 | 1513 | return retval; |
1da177e4 | 1514 | |
8e1611e2 | 1515 | return sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1516 | } |
1517 | ||
9d6a15c3 DB |
1518 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1519 | { | |
1520 | return __sys_socket(family, type, protocol); | |
1521 | } | |
1522 | ||
1da177e4 LT |
1523 | /* |
1524 | * Create a pair of connected sockets. | |
1525 | */ | |
1526 | ||
6debc8d8 | 1527 | int __sys_socketpair(int family, int type, int protocol, int __user *usockvec) |
1da177e4 LT |
1528 | { |
1529 | struct socket *sock1, *sock2; | |
1530 | int fd1, fd2, err; | |
db349509 | 1531 | struct file *newfile1, *newfile2; |
a677a039 UD |
1532 | int flags; |
1533 | ||
1534 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1535 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1536 | return -EINVAL; |
1537 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1538 | |
aaca0bdc UD |
1539 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1540 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1541 | ||
016a266b AV |
1542 | /* |
1543 | * reserve descriptors and make sure we won't fail | |
1544 | * to return them to userland. | |
1545 | */ | |
1546 | fd1 = get_unused_fd_flags(flags); | |
1547 | if (unlikely(fd1 < 0)) | |
1548 | return fd1; | |
1549 | ||
1550 | fd2 = get_unused_fd_flags(flags); | |
1551 | if (unlikely(fd2 < 0)) { | |
1552 | put_unused_fd(fd1); | |
1553 | return fd2; | |
1554 | } | |
1555 | ||
1556 | err = put_user(fd1, &usockvec[0]); | |
1557 | if (err) | |
1558 | goto out; | |
1559 | ||
1560 | err = put_user(fd2, &usockvec[1]); | |
1561 | if (err) | |
1562 | goto out; | |
1563 | ||
1da177e4 LT |
1564 | /* |
1565 | * Obtain the first socket and check if the underlying protocol | |
1566 | * supports the socketpair call. | |
1567 | */ | |
1568 | ||
1569 | err = sock_create(family, type, protocol, &sock1); | |
016a266b | 1570 | if (unlikely(err < 0)) |
1da177e4 LT |
1571 | goto out; |
1572 | ||
1573 | err = sock_create(family, type, protocol, &sock2); | |
016a266b AV |
1574 | if (unlikely(err < 0)) { |
1575 | sock_release(sock1); | |
1576 | goto out; | |
bf3c23d1 | 1577 | } |
d73aa286 | 1578 | |
d47cd945 DH |
1579 | err = security_socket_socketpair(sock1, sock2); |
1580 | if (unlikely(err)) { | |
1581 | sock_release(sock2); | |
1582 | sock_release(sock1); | |
1583 | goto out; | |
1584 | } | |
1585 | ||
016a266b AV |
1586 | err = sock1->ops->socketpair(sock1, sock2); |
1587 | if (unlikely(err < 0)) { | |
1588 | sock_release(sock2); | |
1589 | sock_release(sock1); | |
1590 | goto out; | |
28407630 AV |
1591 | } |
1592 | ||
aab174f0 | 1593 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1594 | if (IS_ERR(newfile1)) { |
28407630 | 1595 | err = PTR_ERR(newfile1); |
016a266b AV |
1596 | sock_release(sock2); |
1597 | goto out; | |
28407630 AV |
1598 | } |
1599 | ||
aab174f0 | 1600 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1601 | if (IS_ERR(newfile2)) { |
1602 | err = PTR_ERR(newfile2); | |
016a266b AV |
1603 | fput(newfile1); |
1604 | goto out; | |
db349509 AV |
1605 | } |
1606 | ||
157cf649 | 1607 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1608 | |
db349509 AV |
1609 | fd_install(fd1, newfile1); |
1610 | fd_install(fd2, newfile2); | |
d73aa286 | 1611 | return 0; |
1da177e4 | 1612 | |
016a266b | 1613 | out: |
d73aa286 | 1614 | put_unused_fd(fd2); |
d73aa286 | 1615 | put_unused_fd(fd1); |
1da177e4 LT |
1616 | return err; |
1617 | } | |
1618 | ||
6debc8d8 DB |
1619 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1620 | int __user *, usockvec) | |
1621 | { | |
1622 | return __sys_socketpair(family, type, protocol, usockvec); | |
1623 | } | |
1624 | ||
1da177e4 LT |
1625 | /* |
1626 | * Bind a name to a socket. Nothing much to do here since it's | |
1627 | * the protocol's responsibility to handle the local address. | |
1628 | * | |
1629 | * We move the socket address to kernel space before we call | |
1630 | * the protocol layer (having also checked the address is ok). | |
1631 | */ | |
1632 | ||
a87d35d8 | 1633 | int __sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
1da177e4 LT |
1634 | { |
1635 | struct socket *sock; | |
230b1839 | 1636 | struct sockaddr_storage address; |
6cb153ca | 1637 | int err, fput_needed; |
1da177e4 | 1638 | |
89bddce5 | 1639 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1640 | if (sock) { |
43db362d | 1641 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
068b88cc | 1642 | if (!err) { |
89bddce5 | 1643 | err = security_socket_bind(sock, |
230b1839 | 1644 | (struct sockaddr *)&address, |
89bddce5 | 1645 | addrlen); |
6cb153ca BL |
1646 | if (!err) |
1647 | err = sock->ops->bind(sock, | |
89bddce5 | 1648 | (struct sockaddr *) |
230b1839 | 1649 | &address, addrlen); |
1da177e4 | 1650 | } |
6cb153ca | 1651 | fput_light(sock->file, fput_needed); |
89bddce5 | 1652 | } |
1da177e4 LT |
1653 | return err; |
1654 | } | |
1655 | ||
a87d35d8 DB |
1656 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1657 | { | |
1658 | return __sys_bind(fd, umyaddr, addrlen); | |
1659 | } | |
1660 | ||
1da177e4 LT |
1661 | /* |
1662 | * Perform a listen. Basically, we allow the protocol to do anything | |
1663 | * necessary for a listen, and if that works, we mark the socket as | |
1664 | * ready for listening. | |
1665 | */ | |
1666 | ||
25e290ee | 1667 | int __sys_listen(int fd, int backlog) |
1da177e4 LT |
1668 | { |
1669 | struct socket *sock; | |
6cb153ca | 1670 | int err, fput_needed; |
b8e1f9b5 | 1671 | int somaxconn; |
89bddce5 SH |
1672 | |
1673 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1674 | if (sock) { | |
8efa6e93 | 1675 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1676 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1677 | backlog = somaxconn; |
1da177e4 LT |
1678 | |
1679 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1680 | if (!err) |
1681 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1682 | |
6cb153ca | 1683 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1684 | } |
1685 | return err; | |
1686 | } | |
1687 | ||
25e290ee DB |
1688 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1689 | { | |
1690 | return __sys_listen(fd, backlog); | |
1691 | } | |
1692 | ||
1da177e4 LT |
1693 | /* |
1694 | * For accept, we attempt to create a new socket, set up the link | |
1695 | * with the client, wake up the client, then return the new | |
1696 | * connected fd. We collect the address of the connector in kernel | |
1697 | * space and move it to user at the very end. This is unclean because | |
1698 | * we open the socket then return an error. | |
1699 | * | |
1700 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1701 | * status to recvmsg. We need to add that support in a way thats | |
b903036a | 1702 | * clean when we restructure accept also. |
1da177e4 LT |
1703 | */ |
1704 | ||
4541e805 DB |
1705 | int __sys_accept4(int fd, struct sockaddr __user *upeer_sockaddr, |
1706 | int __user *upeer_addrlen, int flags) | |
1da177e4 LT |
1707 | { |
1708 | struct socket *sock, *newsock; | |
39d8c1b6 | 1709 | struct file *newfile; |
6cb153ca | 1710 | int err, len, newfd, fput_needed; |
230b1839 | 1711 | struct sockaddr_storage address; |
1da177e4 | 1712 | |
77d27200 | 1713 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1714 | return -EINVAL; |
1715 | ||
1716 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1717 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1718 | ||
6cb153ca | 1719 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1720 | if (!sock) |
1721 | goto out; | |
1722 | ||
1723 | err = -ENFILE; | |
c6d409cf ED |
1724 | newsock = sock_alloc(); |
1725 | if (!newsock) | |
1da177e4 LT |
1726 | goto out_put; |
1727 | ||
1728 | newsock->type = sock->type; | |
1729 | newsock->ops = sock->ops; | |
1730 | ||
1da177e4 LT |
1731 | /* |
1732 | * We don't need try_module_get here, as the listening socket (sock) | |
1733 | * has the protocol module (sock->ops->owner) held. | |
1734 | */ | |
1735 | __module_get(newsock->ops->owner); | |
1736 | ||
28407630 | 1737 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1738 | if (unlikely(newfd < 0)) { |
1739 | err = newfd; | |
9a1875e6 DM |
1740 | sock_release(newsock); |
1741 | goto out_put; | |
39d8c1b6 | 1742 | } |
aab174f0 | 1743 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1744 | if (IS_ERR(newfile)) { |
28407630 AV |
1745 | err = PTR_ERR(newfile); |
1746 | put_unused_fd(newfd); | |
28407630 AV |
1747 | goto out_put; |
1748 | } | |
39d8c1b6 | 1749 | |
a79af59e FF |
1750 | err = security_socket_accept(sock, newsock); |
1751 | if (err) | |
39d8c1b6 | 1752 | goto out_fd; |
a79af59e | 1753 | |
cdfbabfb | 1754 | err = sock->ops->accept(sock, newsock, sock->file->f_flags, false); |
1da177e4 | 1755 | if (err < 0) |
39d8c1b6 | 1756 | goto out_fd; |
1da177e4 LT |
1757 | |
1758 | if (upeer_sockaddr) { | |
9b2c45d4 DV |
1759 | len = newsock->ops->getname(newsock, |
1760 | (struct sockaddr *)&address, 2); | |
1761 | if (len < 0) { | |
1da177e4 | 1762 | err = -ECONNABORTED; |
39d8c1b6 | 1763 | goto out_fd; |
1da177e4 | 1764 | } |
43db362d | 1765 | err = move_addr_to_user(&address, |
230b1839 | 1766 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1767 | if (err < 0) |
39d8c1b6 | 1768 | goto out_fd; |
1da177e4 LT |
1769 | } |
1770 | ||
1771 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1772 | ||
39d8c1b6 DM |
1773 | fd_install(newfd, newfile); |
1774 | err = newfd; | |
1da177e4 | 1775 | |
1da177e4 | 1776 | out_put: |
6cb153ca | 1777 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1778 | out: |
1779 | return err; | |
39d8c1b6 | 1780 | out_fd: |
9606a216 | 1781 | fput(newfile); |
39d8c1b6 | 1782 | put_unused_fd(newfd); |
1da177e4 LT |
1783 | goto out_put; |
1784 | } | |
1785 | ||
4541e805 DB |
1786 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1787 | int __user *, upeer_addrlen, int, flags) | |
1788 | { | |
1789 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, flags); | |
1790 | } | |
1791 | ||
20f37034 HC |
1792 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1793 | int __user *, upeer_addrlen) | |
aaca0bdc | 1794 | { |
4541e805 | 1795 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1796 | } |
1797 | ||
1da177e4 LT |
1798 | /* |
1799 | * Attempt to connect to a socket with the server address. The address | |
1800 | * is in user space so we verify it is OK and move it to kernel space. | |
1801 | * | |
1802 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1803 | * break bindings | |
1804 | * | |
1805 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1806 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1807 | * include the -EINPROGRESS status for such sockets. | |
1808 | */ | |
1809 | ||
1387c2c2 | 1810 | int __sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
1da177e4 LT |
1811 | { |
1812 | struct socket *sock; | |
230b1839 | 1813 | struct sockaddr_storage address; |
6cb153ca | 1814 | int err, fput_needed; |
1da177e4 | 1815 | |
6cb153ca | 1816 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1817 | if (!sock) |
1818 | goto out; | |
43db362d | 1819 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1820 | if (err < 0) |
1821 | goto out_put; | |
1822 | ||
89bddce5 | 1823 | err = |
230b1839 | 1824 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1825 | if (err) |
1826 | goto out_put; | |
1827 | ||
230b1839 | 1828 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1829 | sock->file->f_flags); |
1830 | out_put: | |
6cb153ca | 1831 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1832 | out: |
1833 | return err; | |
1834 | } | |
1835 | ||
1387c2c2 DB |
1836 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1837 | int, addrlen) | |
1838 | { | |
1839 | return __sys_connect(fd, uservaddr, addrlen); | |
1840 | } | |
1841 | ||
1da177e4 LT |
1842 | /* |
1843 | * Get the local address ('name') of a socket object. Move the obtained | |
1844 | * name to user space. | |
1845 | */ | |
1846 | ||
8882a107 DB |
1847 | int __sys_getsockname(int fd, struct sockaddr __user *usockaddr, |
1848 | int __user *usockaddr_len) | |
1da177e4 LT |
1849 | { |
1850 | struct socket *sock; | |
230b1839 | 1851 | struct sockaddr_storage address; |
9b2c45d4 | 1852 | int err, fput_needed; |
89bddce5 | 1853 | |
6cb153ca | 1854 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1855 | if (!sock) |
1856 | goto out; | |
1857 | ||
1858 | err = security_socket_getsockname(sock); | |
1859 | if (err) | |
1860 | goto out_put; | |
1861 | ||
9b2c45d4 DV |
1862 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 0); |
1863 | if (err < 0) | |
1da177e4 | 1864 | goto out_put; |
9b2c45d4 DV |
1865 | /* "err" is actually length in this case */ |
1866 | err = move_addr_to_user(&address, err, usockaddr, usockaddr_len); | |
1da177e4 LT |
1867 | |
1868 | out_put: | |
6cb153ca | 1869 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1870 | out: |
1871 | return err; | |
1872 | } | |
1873 | ||
8882a107 DB |
1874 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1875 | int __user *, usockaddr_len) | |
1876 | { | |
1877 | return __sys_getsockname(fd, usockaddr, usockaddr_len); | |
1878 | } | |
1879 | ||
1da177e4 LT |
1880 | /* |
1881 | * Get the remote address ('name') of a socket object. Move the obtained | |
1882 | * name to user space. | |
1883 | */ | |
1884 | ||
b21c8f83 DB |
1885 | int __sys_getpeername(int fd, struct sockaddr __user *usockaddr, |
1886 | int __user *usockaddr_len) | |
1da177e4 LT |
1887 | { |
1888 | struct socket *sock; | |
230b1839 | 1889 | struct sockaddr_storage address; |
9b2c45d4 | 1890 | int err, fput_needed; |
1da177e4 | 1891 | |
89bddce5 SH |
1892 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1893 | if (sock != NULL) { | |
1da177e4 LT |
1894 | err = security_socket_getpeername(sock); |
1895 | if (err) { | |
6cb153ca | 1896 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1897 | return err; |
1898 | } | |
1899 | ||
9b2c45d4 DV |
1900 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 1); |
1901 | if (err >= 0) | |
1902 | /* "err" is actually length in this case */ | |
1903 | err = move_addr_to_user(&address, err, usockaddr, | |
89bddce5 | 1904 | usockaddr_len); |
6cb153ca | 1905 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1906 | } |
1907 | return err; | |
1908 | } | |
1909 | ||
b21c8f83 DB |
1910 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1911 | int __user *, usockaddr_len) | |
1912 | { | |
1913 | return __sys_getpeername(fd, usockaddr, usockaddr_len); | |
1914 | } | |
1915 | ||
1da177e4 LT |
1916 | /* |
1917 | * Send a datagram to a given address. We move the address into kernel | |
1918 | * space and check the user space data area is readable before invoking | |
1919 | * the protocol. | |
1920 | */ | |
211b634b DB |
1921 | int __sys_sendto(int fd, void __user *buff, size_t len, unsigned int flags, |
1922 | struct sockaddr __user *addr, int addr_len) | |
1da177e4 LT |
1923 | { |
1924 | struct socket *sock; | |
230b1839 | 1925 | struct sockaddr_storage address; |
1da177e4 LT |
1926 | int err; |
1927 | struct msghdr msg; | |
1928 | struct iovec iov; | |
6cb153ca | 1929 | int fput_needed; |
6cb153ca | 1930 | |
602bd0e9 AV |
1931 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1932 | if (unlikely(err)) | |
1933 | return err; | |
de0fa95c PE |
1934 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1935 | if (!sock) | |
4387ff75 | 1936 | goto out; |
6cb153ca | 1937 | |
89bddce5 | 1938 | msg.msg_name = NULL; |
89bddce5 SH |
1939 | msg.msg_control = NULL; |
1940 | msg.msg_controllen = 0; | |
1941 | msg.msg_namelen = 0; | |
6cb153ca | 1942 | if (addr) { |
43db362d | 1943 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1944 | if (err < 0) |
1945 | goto out_put; | |
230b1839 | 1946 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1947 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1948 | } |
1949 | if (sock->file->f_flags & O_NONBLOCK) | |
1950 | flags |= MSG_DONTWAIT; | |
1951 | msg.msg_flags = flags; | |
d8725c86 | 1952 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 1953 | |
89bddce5 | 1954 | out_put: |
de0fa95c | 1955 | fput_light(sock->file, fput_needed); |
4387ff75 | 1956 | out: |
1da177e4 LT |
1957 | return err; |
1958 | } | |
1959 | ||
211b634b DB |
1960 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
1961 | unsigned int, flags, struct sockaddr __user *, addr, | |
1962 | int, addr_len) | |
1963 | { | |
1964 | return __sys_sendto(fd, buff, len, flags, addr, addr_len); | |
1965 | } | |
1966 | ||
1da177e4 | 1967 | /* |
89bddce5 | 1968 | * Send a datagram down a socket. |
1da177e4 LT |
1969 | */ |
1970 | ||
3e0fa65f | 1971 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1972 | unsigned int, flags) |
1da177e4 | 1973 | { |
211b634b | 1974 | return __sys_sendto(fd, buff, len, flags, NULL, 0); |
1da177e4 LT |
1975 | } |
1976 | ||
1977 | /* | |
89bddce5 | 1978 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1979 | * sender. We verify the buffers are writable and if needed move the |
1980 | * sender address from kernel to user space. | |
1981 | */ | |
7a09e1eb DB |
1982 | int __sys_recvfrom(int fd, void __user *ubuf, size_t size, unsigned int flags, |
1983 | struct sockaddr __user *addr, int __user *addr_len) | |
1da177e4 LT |
1984 | { |
1985 | struct socket *sock; | |
1986 | struct iovec iov; | |
1987 | struct msghdr msg; | |
230b1839 | 1988 | struct sockaddr_storage address; |
89bddce5 | 1989 | int err, err2; |
6cb153ca BL |
1990 | int fput_needed; |
1991 | ||
602bd0e9 AV |
1992 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
1993 | if (unlikely(err)) | |
1994 | return err; | |
de0fa95c | 1995 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 1996 | if (!sock) |
de0fa95c | 1997 | goto out; |
1da177e4 | 1998 | |
89bddce5 SH |
1999 | msg.msg_control = NULL; |
2000 | msg.msg_controllen = 0; | |
f3d33426 HFS |
2001 | /* Save some cycles and don't copy the address if not needed */ |
2002 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
2003 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
2004 | msg.msg_namelen = 0; | |
130ed5d1 | 2005 | msg.msg_iocb = NULL; |
9f138fa6 | 2006 | msg.msg_flags = 0; |
1da177e4 LT |
2007 | if (sock->file->f_flags & O_NONBLOCK) |
2008 | flags |= MSG_DONTWAIT; | |
2da62906 | 2009 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 2010 | |
89bddce5 | 2011 | if (err >= 0 && addr != NULL) { |
43db362d | 2012 | err2 = move_addr_to_user(&address, |
230b1839 | 2013 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
2014 | if (err2 < 0) |
2015 | err = err2; | |
1da177e4 | 2016 | } |
de0fa95c PE |
2017 | |
2018 | fput_light(sock->file, fput_needed); | |
4387ff75 | 2019 | out: |
1da177e4 LT |
2020 | return err; |
2021 | } | |
2022 | ||
7a09e1eb DB |
2023 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
2024 | unsigned int, flags, struct sockaddr __user *, addr, | |
2025 | int __user *, addr_len) | |
2026 | { | |
2027 | return __sys_recvfrom(fd, ubuf, size, flags, addr, addr_len); | |
2028 | } | |
2029 | ||
1da177e4 | 2030 | /* |
89bddce5 | 2031 | * Receive a datagram from a socket. |
1da177e4 LT |
2032 | */ |
2033 | ||
b7c0ddf5 JG |
2034 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
2035 | unsigned int, flags) | |
1da177e4 | 2036 | { |
7a09e1eb | 2037 | return __sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
1da177e4 LT |
2038 | } |
2039 | ||
2040 | /* | |
2041 | * Set a socket option. Because we don't know the option lengths we have | |
2042 | * to pass the user mode parameter for the protocols to sort out. | |
2043 | */ | |
2044 | ||
cc36dca0 DB |
2045 | static int __sys_setsockopt(int fd, int level, int optname, |
2046 | char __user *optval, int optlen) | |
1da177e4 | 2047 | { |
0d01da6a SF |
2048 | mm_segment_t oldfs = get_fs(); |
2049 | char *kernel_optval = NULL; | |
6cb153ca | 2050 | int err, fput_needed; |
1da177e4 LT |
2051 | struct socket *sock; |
2052 | ||
2053 | if (optlen < 0) | |
2054 | return -EINVAL; | |
89bddce5 SH |
2055 | |
2056 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2057 | if (sock != NULL) { | |
2058 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
2059 | if (err) |
2060 | goto out_put; | |
1da177e4 | 2061 | |
0d01da6a SF |
2062 | err = BPF_CGROUP_RUN_PROG_SETSOCKOPT(sock->sk, &level, |
2063 | &optname, optval, &optlen, | |
2064 | &kernel_optval); | |
2065 | ||
2066 | if (err < 0) { | |
2067 | goto out_put; | |
2068 | } else if (err > 0) { | |
2069 | err = 0; | |
2070 | goto out_put; | |
2071 | } | |
2072 | ||
2073 | if (kernel_optval) { | |
2074 | set_fs(KERNEL_DS); | |
2075 | optval = (char __user __force *)kernel_optval; | |
2076 | } | |
2077 | ||
1da177e4 | 2078 | if (level == SOL_SOCKET) |
89bddce5 SH |
2079 | err = |
2080 | sock_setsockopt(sock, level, optname, optval, | |
2081 | optlen); | |
1da177e4 | 2082 | else |
89bddce5 SH |
2083 | err = |
2084 | sock->ops->setsockopt(sock, level, optname, optval, | |
2085 | optlen); | |
0d01da6a SF |
2086 | |
2087 | if (kernel_optval) { | |
2088 | set_fs(oldfs); | |
2089 | kfree(kernel_optval); | |
2090 | } | |
6cb153ca BL |
2091 | out_put: |
2092 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2093 | } |
2094 | return err; | |
2095 | } | |
2096 | ||
cc36dca0 DB |
2097 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
2098 | char __user *, optval, int, optlen) | |
2099 | { | |
2100 | return __sys_setsockopt(fd, level, optname, optval, optlen); | |
2101 | } | |
2102 | ||
1da177e4 LT |
2103 | /* |
2104 | * Get a socket option. Because we don't know the option lengths we have | |
2105 | * to pass a user mode parameter for the protocols to sort out. | |
2106 | */ | |
2107 | ||
13a2d70e DB |
2108 | static int __sys_getsockopt(int fd, int level, int optname, |
2109 | char __user *optval, int __user *optlen) | |
1da177e4 | 2110 | { |
6cb153ca | 2111 | int err, fput_needed; |
1da177e4 | 2112 | struct socket *sock; |
0d01da6a | 2113 | int max_optlen; |
1da177e4 | 2114 | |
89bddce5 SH |
2115 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2116 | if (sock != NULL) { | |
6cb153ca BL |
2117 | err = security_socket_getsockopt(sock, level, optname); |
2118 | if (err) | |
2119 | goto out_put; | |
1da177e4 | 2120 | |
0d01da6a SF |
2121 | max_optlen = BPF_CGROUP_GETSOCKOPT_MAX_OPTLEN(optlen); |
2122 | ||
1da177e4 | 2123 | if (level == SOL_SOCKET) |
89bddce5 SH |
2124 | err = |
2125 | sock_getsockopt(sock, level, optname, optval, | |
2126 | optlen); | |
1da177e4 | 2127 | else |
89bddce5 SH |
2128 | err = |
2129 | sock->ops->getsockopt(sock, level, optname, optval, | |
2130 | optlen); | |
0d01da6a SF |
2131 | |
2132 | err = BPF_CGROUP_RUN_PROG_GETSOCKOPT(sock->sk, level, optname, | |
2133 | optval, optlen, | |
2134 | max_optlen, err); | |
6cb153ca BL |
2135 | out_put: |
2136 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2137 | } |
2138 | return err; | |
2139 | } | |
2140 | ||
13a2d70e DB |
2141 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
2142 | char __user *, optval, int __user *, optlen) | |
2143 | { | |
2144 | return __sys_getsockopt(fd, level, optname, optval, optlen); | |
2145 | } | |
2146 | ||
1da177e4 LT |
2147 | /* |
2148 | * Shutdown a socket. | |
2149 | */ | |
2150 | ||
005a1aea | 2151 | int __sys_shutdown(int fd, int how) |
1da177e4 | 2152 | { |
6cb153ca | 2153 | int err, fput_needed; |
1da177e4 LT |
2154 | struct socket *sock; |
2155 | ||
89bddce5 SH |
2156 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2157 | if (sock != NULL) { | |
1da177e4 | 2158 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
2159 | if (!err) |
2160 | err = sock->ops->shutdown(sock, how); | |
2161 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2162 | } |
2163 | return err; | |
2164 | } | |
2165 | ||
005a1aea DB |
2166 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
2167 | { | |
2168 | return __sys_shutdown(fd, how); | |
2169 | } | |
2170 | ||
89bddce5 | 2171 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
2172 | * fields which are the same type (int / unsigned) on our platforms. |
2173 | */ | |
2174 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
2175 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
2176 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
2177 | ||
c71d8ebe TH |
2178 | struct used_address { |
2179 | struct sockaddr_storage name; | |
2180 | unsigned int name_len; | |
2181 | }; | |
2182 | ||
da184284 AV |
2183 | static int copy_msghdr_from_user(struct msghdr *kmsg, |
2184 | struct user_msghdr __user *umsg, | |
2185 | struct sockaddr __user **save_addr, | |
2186 | struct iovec **iov) | |
1661bf36 | 2187 | { |
ffb07550 | 2188 | struct user_msghdr msg; |
08adb7da AV |
2189 | ssize_t err; |
2190 | ||
ffb07550 | 2191 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 2192 | return -EFAULT; |
dbb490b9 | 2193 | |
864d9664 | 2194 | kmsg->msg_control = (void __force *)msg.msg_control; |
ffb07550 AV |
2195 | kmsg->msg_controllen = msg.msg_controllen; |
2196 | kmsg->msg_flags = msg.msg_flags; | |
2197 | ||
2198 | kmsg->msg_namelen = msg.msg_namelen; | |
2199 | if (!msg.msg_name) | |
6a2a2b3a AS |
2200 | kmsg->msg_namelen = 0; |
2201 | ||
dbb490b9 ML |
2202 | if (kmsg->msg_namelen < 0) |
2203 | return -EINVAL; | |
2204 | ||
1661bf36 | 2205 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 2206 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
2207 | |
2208 | if (save_addr) | |
ffb07550 | 2209 | *save_addr = msg.msg_name; |
08adb7da | 2210 | |
ffb07550 | 2211 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 2212 | if (!save_addr) { |
864d9664 PA |
2213 | err = move_addr_to_kernel(msg.msg_name, |
2214 | kmsg->msg_namelen, | |
08adb7da AV |
2215 | kmsg->msg_name); |
2216 | if (err < 0) | |
2217 | return err; | |
2218 | } | |
2219 | } else { | |
2220 | kmsg->msg_name = NULL; | |
2221 | kmsg->msg_namelen = 0; | |
2222 | } | |
2223 | ||
ffb07550 | 2224 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
2225 | return -EMSGSIZE; |
2226 | ||
0345f931 | 2227 | kmsg->msg_iocb = NULL; |
2228 | ||
87e5e6da | 2229 | err = import_iovec(save_addr ? READ : WRITE, |
ffb07550 | 2230 | msg.msg_iov, msg.msg_iovlen, |
da184284 | 2231 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
87e5e6da | 2232 | return err < 0 ? err : 0; |
1661bf36 DC |
2233 | } |
2234 | ||
666547ff | 2235 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2236 | struct msghdr *msg_sys, unsigned int flags, |
28a94d8f TH |
2237 | struct used_address *used_address, |
2238 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 2239 | { |
89bddce5 SH |
2240 | struct compat_msghdr __user *msg_compat = |
2241 | (struct compat_msghdr __user *)msg; | |
230b1839 | 2242 | struct sockaddr_storage address; |
1da177e4 | 2243 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 2244 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 2245 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 2246 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 2247 | unsigned char *ctl_buf = ctl; |
d8725c86 | 2248 | int ctl_len; |
08adb7da | 2249 | ssize_t err; |
89bddce5 | 2250 | |
08adb7da | 2251 | msg_sys->msg_name = &address; |
1da177e4 | 2252 | |
08449320 | 2253 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2254 | err = get_compat_msghdr(msg_sys, msg_compat, NULL, &iov); |
08449320 | 2255 | else |
08adb7da | 2256 | err = copy_msghdr_from_user(msg_sys, msg, NULL, &iov); |
89bddce5 | 2257 | if (err < 0) |
da184284 | 2258 | return err; |
1da177e4 LT |
2259 | |
2260 | err = -ENOBUFS; | |
2261 | ||
228e548e | 2262 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 2263 | goto out_freeiov; |
28a94d8f | 2264 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2265 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2266 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2267 | err = |
228e548e | 2268 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2269 | sizeof(ctl)); |
1da177e4 LT |
2270 | if (err) |
2271 | goto out_freeiov; | |
228e548e AB |
2272 | ctl_buf = msg_sys->msg_control; |
2273 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2274 | } else if (ctl_len) { |
ac4340fc DM |
2275 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2276 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2277 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2278 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2279 | if (ctl_buf == NULL) |
1da177e4 LT |
2280 | goto out_freeiov; |
2281 | } | |
2282 | err = -EFAULT; | |
2283 | /* | |
228e548e | 2284 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2285 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2286 | * checking falls down on this. | |
2287 | */ | |
fb8621bb | 2288 | if (copy_from_user(ctl_buf, |
228e548e | 2289 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2290 | ctl_len)) |
1da177e4 | 2291 | goto out_freectl; |
228e548e | 2292 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2293 | } |
228e548e | 2294 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2295 | |
2296 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2297 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2298 | /* |
2299 | * If this is sendmmsg() and current destination address is same as | |
2300 | * previously succeeded address, omit asking LSM's decision. | |
2301 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2302 | * destination address never matches. | |
2303 | */ | |
bc909d9d MD |
2304 | if (used_address && msg_sys->msg_name && |
2305 | used_address->name_len == msg_sys->msg_namelen && | |
2306 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2307 | used_address->name_len)) { |
d8725c86 | 2308 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2309 | goto out_freectl; |
2310 | } | |
d8725c86 | 2311 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2312 | /* |
2313 | * If this is sendmmsg() and sending to current destination address was | |
2314 | * successful, remember it. | |
2315 | */ | |
2316 | if (used_address && err >= 0) { | |
2317 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2318 | if (msg_sys->msg_name) |
2319 | memcpy(&used_address->name, msg_sys->msg_name, | |
2320 | used_address->name_len); | |
c71d8ebe | 2321 | } |
1da177e4 LT |
2322 | |
2323 | out_freectl: | |
89bddce5 | 2324 | if (ctl_buf != ctl) |
1da177e4 LT |
2325 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
2326 | out_freeiov: | |
da184284 | 2327 | kfree(iov); |
228e548e AB |
2328 | return err; |
2329 | } | |
2330 | ||
2331 | /* | |
2332 | * BSD sendmsg interface | |
2333 | */ | |
0fa03c62 JA |
2334 | long __sys_sendmsg_sock(struct socket *sock, struct user_msghdr __user *msg, |
2335 | unsigned int flags) | |
2336 | { | |
2337 | struct msghdr msg_sys; | |
2338 | ||
2339 | return ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); | |
2340 | } | |
228e548e | 2341 | |
e1834a32 DB |
2342 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2343 | bool forbid_cmsg_compat) | |
228e548e AB |
2344 | { |
2345 | int fput_needed, err; | |
2346 | struct msghdr msg_sys; | |
1be374a0 AL |
2347 | struct socket *sock; |
2348 | ||
e1834a32 DB |
2349 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2350 | return -EINVAL; | |
2351 | ||
1be374a0 | 2352 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2353 | if (!sock) |
2354 | goto out; | |
2355 | ||
28a94d8f | 2356 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2357 | |
6cb153ca | 2358 | fput_light(sock->file, fput_needed); |
89bddce5 | 2359 | out: |
1da177e4 LT |
2360 | return err; |
2361 | } | |
2362 | ||
666547ff | 2363 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 | 2364 | { |
e1834a32 | 2365 | return __sys_sendmsg(fd, msg, flags, true); |
a7526eb5 AL |
2366 | } |
2367 | ||
228e548e AB |
2368 | /* |
2369 | * Linux sendmmsg interface | |
2370 | */ | |
2371 | ||
2372 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
e1834a32 | 2373 | unsigned int flags, bool forbid_cmsg_compat) |
228e548e AB |
2374 | { |
2375 | int fput_needed, err, datagrams; | |
2376 | struct socket *sock; | |
2377 | struct mmsghdr __user *entry; | |
2378 | struct compat_mmsghdr __user *compat_entry; | |
2379 | struct msghdr msg_sys; | |
c71d8ebe | 2380 | struct used_address used_address; |
f092276d | 2381 | unsigned int oflags = flags; |
228e548e | 2382 | |
e1834a32 DB |
2383 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2384 | return -EINVAL; | |
2385 | ||
98382f41 AB |
2386 | if (vlen > UIO_MAXIOV) |
2387 | vlen = UIO_MAXIOV; | |
228e548e AB |
2388 | |
2389 | datagrams = 0; | |
2390 | ||
2391 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2392 | if (!sock) | |
2393 | return err; | |
2394 | ||
c71d8ebe | 2395 | used_address.name_len = UINT_MAX; |
228e548e AB |
2396 | entry = mmsg; |
2397 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2398 | err = 0; |
f092276d | 2399 | flags |= MSG_BATCH; |
228e548e AB |
2400 | |
2401 | while (datagrams < vlen) { | |
f092276d TH |
2402 | if (datagrams == vlen - 1) |
2403 | flags = oflags; | |
2404 | ||
228e548e | 2405 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2406 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2407 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2408 | if (err < 0) |
2409 | break; | |
2410 | err = __put_user(err, &compat_entry->msg_len); | |
2411 | ++compat_entry; | |
2412 | } else { | |
a7526eb5 | 2413 | err = ___sys_sendmsg(sock, |
666547ff | 2414 | (struct user_msghdr __user *)entry, |
28a94d8f | 2415 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2416 | if (err < 0) |
2417 | break; | |
2418 | err = put_user(err, &entry->msg_len); | |
2419 | ++entry; | |
2420 | } | |
2421 | ||
2422 | if (err) | |
2423 | break; | |
2424 | ++datagrams; | |
3023898b SHY |
2425 | if (msg_data_left(&msg_sys)) |
2426 | break; | |
a78cb84c | 2427 | cond_resched(); |
228e548e AB |
2428 | } |
2429 | ||
228e548e AB |
2430 | fput_light(sock->file, fput_needed); |
2431 | ||
728ffb86 AB |
2432 | /* We only return an error if no datagrams were able to be sent */ |
2433 | if (datagrams != 0) | |
228e548e AB |
2434 | return datagrams; |
2435 | ||
228e548e AB |
2436 | return err; |
2437 | } | |
2438 | ||
2439 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2440 | unsigned int, vlen, unsigned int, flags) | |
2441 | { | |
e1834a32 | 2442 | return __sys_sendmmsg(fd, mmsg, vlen, flags, true); |
228e548e AB |
2443 | } |
2444 | ||
666547ff | 2445 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2446 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2447 | { |
89bddce5 SH |
2448 | struct compat_msghdr __user *msg_compat = |
2449 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2450 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2451 | struct iovec *iov = iovstack; |
1da177e4 | 2452 | unsigned long cmsg_ptr; |
2da62906 | 2453 | int len; |
08adb7da | 2454 | ssize_t err; |
1da177e4 LT |
2455 | |
2456 | /* kernel mode address */ | |
230b1839 | 2457 | struct sockaddr_storage addr; |
1da177e4 LT |
2458 | |
2459 | /* user mode address pointers */ | |
2460 | struct sockaddr __user *uaddr; | |
08adb7da | 2461 | int __user *uaddr_len = COMPAT_NAMELEN(msg); |
89bddce5 | 2462 | |
08adb7da | 2463 | msg_sys->msg_name = &addr; |
1da177e4 | 2464 | |
f3d33426 | 2465 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2466 | err = get_compat_msghdr(msg_sys, msg_compat, &uaddr, &iov); |
f3d33426 | 2467 | else |
08adb7da | 2468 | err = copy_msghdr_from_user(msg_sys, msg, &uaddr, &iov); |
1da177e4 | 2469 | if (err < 0) |
da184284 | 2470 | return err; |
1da177e4 | 2471 | |
a2e27255 ACM |
2472 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2473 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2474 | |
f3d33426 HFS |
2475 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2476 | msg_sys->msg_namelen = 0; | |
2477 | ||
1da177e4 LT |
2478 | if (sock->file->f_flags & O_NONBLOCK) |
2479 | flags |= MSG_DONTWAIT; | |
2da62906 | 2480 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, flags); |
1da177e4 LT |
2481 | if (err < 0) |
2482 | goto out_freeiov; | |
2483 | len = err; | |
2484 | ||
2485 | if (uaddr != NULL) { | |
43db362d | 2486 | err = move_addr_to_user(&addr, |
a2e27255 | 2487 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2488 | uaddr_len); |
1da177e4 LT |
2489 | if (err < 0) |
2490 | goto out_freeiov; | |
2491 | } | |
a2e27255 | 2492 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2493 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2494 | if (err) |
2495 | goto out_freeiov; | |
2496 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2497 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2498 | &msg_compat->msg_controllen); |
2499 | else | |
a2e27255 | 2500 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2501 | &msg->msg_controllen); |
2502 | if (err) | |
2503 | goto out_freeiov; | |
2504 | err = len; | |
2505 | ||
2506 | out_freeiov: | |
da184284 | 2507 | kfree(iov); |
a2e27255 ACM |
2508 | return err; |
2509 | } | |
2510 | ||
2511 | /* | |
2512 | * BSD recvmsg interface | |
2513 | */ | |
2514 | ||
aa1fa28f JA |
2515 | long __sys_recvmsg_sock(struct socket *sock, struct user_msghdr __user *msg, |
2516 | unsigned int flags) | |
2517 | { | |
2518 | struct msghdr msg_sys; | |
2519 | ||
2520 | return ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); | |
2521 | } | |
2522 | ||
e1834a32 DB |
2523 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2524 | bool forbid_cmsg_compat) | |
a2e27255 ACM |
2525 | { |
2526 | int fput_needed, err; | |
2527 | struct msghdr msg_sys; | |
1be374a0 AL |
2528 | struct socket *sock; |
2529 | ||
e1834a32 DB |
2530 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2531 | return -EINVAL; | |
2532 | ||
1be374a0 | 2533 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2534 | if (!sock) |
2535 | goto out; | |
2536 | ||
a7526eb5 | 2537 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2538 | |
6cb153ca | 2539 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2540 | out: |
2541 | return err; | |
2542 | } | |
2543 | ||
666547ff | 2544 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2545 | unsigned int, flags) |
2546 | { | |
e1834a32 | 2547 | return __sys_recvmsg(fd, msg, flags, true); |
a7526eb5 AL |
2548 | } |
2549 | ||
a2e27255 ACM |
2550 | /* |
2551 | * Linux recvmmsg interface | |
2552 | */ | |
2553 | ||
e11d4284 AB |
2554 | static int do_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2555 | unsigned int vlen, unsigned int flags, | |
2556 | struct timespec64 *timeout) | |
a2e27255 ACM |
2557 | { |
2558 | int fput_needed, err, datagrams; | |
2559 | struct socket *sock; | |
2560 | struct mmsghdr __user *entry; | |
d7256d0e | 2561 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2562 | struct msghdr msg_sys; |
766b9f92 DD |
2563 | struct timespec64 end_time; |
2564 | struct timespec64 timeout64; | |
a2e27255 ACM |
2565 | |
2566 | if (timeout && | |
2567 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2568 | timeout->tv_nsec)) | |
2569 | return -EINVAL; | |
2570 | ||
2571 | datagrams = 0; | |
2572 | ||
2573 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2574 | if (!sock) | |
2575 | return err; | |
2576 | ||
7797dc41 SHY |
2577 | if (likely(!(flags & MSG_ERRQUEUE))) { |
2578 | err = sock_error(sock->sk); | |
2579 | if (err) { | |
2580 | datagrams = err; | |
2581 | goto out_put; | |
2582 | } | |
e623a9e9 | 2583 | } |
a2e27255 ACM |
2584 | |
2585 | entry = mmsg; | |
d7256d0e | 2586 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2587 | |
2588 | while (datagrams < vlen) { | |
2589 | /* | |
2590 | * No need to ask LSM for more than the first datagram. | |
2591 | */ | |
d7256d0e | 2592 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2593 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2594 | &msg_sys, flags & ~MSG_WAITFORONE, |
2595 | datagrams); | |
d7256d0e JMG |
2596 | if (err < 0) |
2597 | break; | |
2598 | err = __put_user(err, &compat_entry->msg_len); | |
2599 | ++compat_entry; | |
2600 | } else { | |
a7526eb5 | 2601 | err = ___sys_recvmsg(sock, |
666547ff | 2602 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2603 | &msg_sys, flags & ~MSG_WAITFORONE, |
2604 | datagrams); | |
d7256d0e JMG |
2605 | if (err < 0) |
2606 | break; | |
2607 | err = put_user(err, &entry->msg_len); | |
2608 | ++entry; | |
2609 | } | |
2610 | ||
a2e27255 ACM |
2611 | if (err) |
2612 | break; | |
a2e27255 ACM |
2613 | ++datagrams; |
2614 | ||
71c5c159 BB |
2615 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2616 | if (flags & MSG_WAITFORONE) | |
2617 | flags |= MSG_DONTWAIT; | |
2618 | ||
a2e27255 | 2619 | if (timeout) { |
766b9f92 | 2620 | ktime_get_ts64(&timeout64); |
c2e6c856 | 2621 | *timeout = timespec64_sub(end_time, timeout64); |
a2e27255 ACM |
2622 | if (timeout->tv_sec < 0) { |
2623 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2624 | break; | |
2625 | } | |
2626 | ||
2627 | /* Timeout, return less than vlen datagrams */ | |
2628 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2629 | break; | |
2630 | } | |
2631 | ||
2632 | /* Out of band data, return right away */ | |
2633 | if (msg_sys.msg_flags & MSG_OOB) | |
2634 | break; | |
a78cb84c | 2635 | cond_resched(); |
a2e27255 ACM |
2636 | } |
2637 | ||
a2e27255 | 2638 | if (err == 0) |
34b88a68 ACM |
2639 | goto out_put; |
2640 | ||
2641 | if (datagrams == 0) { | |
2642 | datagrams = err; | |
2643 | goto out_put; | |
2644 | } | |
a2e27255 | 2645 | |
34b88a68 ACM |
2646 | /* |
2647 | * We may return less entries than requested (vlen) if the | |
2648 | * sock is non block and there aren't enough datagrams... | |
2649 | */ | |
2650 | if (err != -EAGAIN) { | |
a2e27255 | 2651 | /* |
34b88a68 ACM |
2652 | * ... or if recvmsg returns an error after we |
2653 | * received some datagrams, where we record the | |
2654 | * error to return on the next call or if the | |
2655 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2656 | */ |
34b88a68 | 2657 | sock->sk->sk_err = -err; |
a2e27255 | 2658 | } |
34b88a68 ACM |
2659 | out_put: |
2660 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2661 | |
34b88a68 | 2662 | return datagrams; |
a2e27255 ACM |
2663 | } |
2664 | ||
e11d4284 AB |
2665 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2666 | unsigned int vlen, unsigned int flags, | |
2667 | struct __kernel_timespec __user *timeout, | |
2668 | struct old_timespec32 __user *timeout32) | |
a2e27255 ACM |
2669 | { |
2670 | int datagrams; | |
c2e6c856 | 2671 | struct timespec64 timeout_sys; |
a2e27255 | 2672 | |
e11d4284 AB |
2673 | if (timeout && get_timespec64(&timeout_sys, timeout)) |
2674 | return -EFAULT; | |
a2e27255 | 2675 | |
e11d4284 | 2676 | if (timeout32 && get_old_timespec32(&timeout_sys, timeout32)) |
a2e27255 ACM |
2677 | return -EFAULT; |
2678 | ||
e11d4284 AB |
2679 | if (!timeout && !timeout32) |
2680 | return do_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2681 | ||
2682 | datagrams = do_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
a2e27255 | 2683 | |
e11d4284 AB |
2684 | if (datagrams <= 0) |
2685 | return datagrams; | |
2686 | ||
2687 | if (timeout && put_timespec64(&timeout_sys, timeout)) | |
2688 | datagrams = -EFAULT; | |
2689 | ||
2690 | if (timeout32 && put_old_timespec32(&timeout_sys, timeout32)) | |
a2e27255 ACM |
2691 | datagrams = -EFAULT; |
2692 | ||
2693 | return datagrams; | |
2694 | } | |
2695 | ||
1255e269 DB |
2696 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, |
2697 | unsigned int, vlen, unsigned int, flags, | |
c2e6c856 | 2698 | struct __kernel_timespec __user *, timeout) |
1255e269 | 2699 | { |
e11d4284 AB |
2700 | if (flags & MSG_CMSG_COMPAT) |
2701 | return -EINVAL; | |
2702 | ||
2703 | return __sys_recvmmsg(fd, mmsg, vlen, flags, timeout, NULL); | |
2704 | } | |
2705 | ||
2706 | #ifdef CONFIG_COMPAT_32BIT_TIME | |
2707 | SYSCALL_DEFINE5(recvmmsg_time32, int, fd, struct mmsghdr __user *, mmsg, | |
2708 | unsigned int, vlen, unsigned int, flags, | |
2709 | struct old_timespec32 __user *, timeout) | |
2710 | { | |
2711 | if (flags & MSG_CMSG_COMPAT) | |
2712 | return -EINVAL; | |
2713 | ||
2714 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL, timeout); | |
1255e269 | 2715 | } |
e11d4284 | 2716 | #endif |
1255e269 | 2717 | |
a2e27255 | 2718 | #ifdef __ARCH_WANT_SYS_SOCKETCALL |
1da177e4 LT |
2719 | /* Argument list sizes for sys_socketcall */ |
2720 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2721 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2722 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2723 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2724 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2725 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2726 | }; |
2727 | ||
1da177e4 LT |
2728 | #undef AL |
2729 | ||
2730 | /* | |
89bddce5 | 2731 | * System call vectors. |
1da177e4 LT |
2732 | * |
2733 | * Argument checking cleaned up. Saved 20% in size. | |
2734 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2735 | * it is set by the callees. |
1da177e4 LT |
2736 | */ |
2737 | ||
3e0fa65f | 2738 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2739 | { |
2950fa9d | 2740 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2741 | unsigned long a0, a1; |
1da177e4 | 2742 | int err; |
47379052 | 2743 | unsigned int len; |
1da177e4 | 2744 | |
228e548e | 2745 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 | 2746 | return -EINVAL; |
c8e8cd57 | 2747 | call = array_index_nospec(call, SYS_SENDMMSG + 1); |
1da177e4 | 2748 | |
47379052 AV |
2749 | len = nargs[call]; |
2750 | if (len > sizeof(a)) | |
2751 | return -EINVAL; | |
2752 | ||
1da177e4 | 2753 | /* copy_from_user should be SMP safe. */ |
47379052 | 2754 | if (copy_from_user(a, args, len)) |
1da177e4 | 2755 | return -EFAULT; |
3ec3b2fb | 2756 | |
2950fa9d CG |
2757 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2758 | if (err) | |
2759 | return err; | |
3ec3b2fb | 2760 | |
89bddce5 SH |
2761 | a0 = a[0]; |
2762 | a1 = a[1]; | |
2763 | ||
2764 | switch (call) { | |
2765 | case SYS_SOCKET: | |
9d6a15c3 | 2766 | err = __sys_socket(a0, a1, a[2]); |
89bddce5 SH |
2767 | break; |
2768 | case SYS_BIND: | |
a87d35d8 | 2769 | err = __sys_bind(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2770 | break; |
2771 | case SYS_CONNECT: | |
1387c2c2 | 2772 | err = __sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2773 | break; |
2774 | case SYS_LISTEN: | |
25e290ee | 2775 | err = __sys_listen(a0, a1); |
89bddce5 SH |
2776 | break; |
2777 | case SYS_ACCEPT: | |
4541e805 DB |
2778 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2779 | (int __user *)a[2], 0); | |
89bddce5 SH |
2780 | break; |
2781 | case SYS_GETSOCKNAME: | |
2782 | err = | |
8882a107 DB |
2783 | __sys_getsockname(a0, (struct sockaddr __user *)a1, |
2784 | (int __user *)a[2]); | |
89bddce5 SH |
2785 | break; |
2786 | case SYS_GETPEERNAME: | |
2787 | err = | |
b21c8f83 DB |
2788 | __sys_getpeername(a0, (struct sockaddr __user *)a1, |
2789 | (int __user *)a[2]); | |
89bddce5 SH |
2790 | break; |
2791 | case SYS_SOCKETPAIR: | |
6debc8d8 | 2792 | err = __sys_socketpair(a0, a1, a[2], (int __user *)a[3]); |
89bddce5 SH |
2793 | break; |
2794 | case SYS_SEND: | |
f3bf896b DB |
2795 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2796 | NULL, 0); | |
89bddce5 SH |
2797 | break; |
2798 | case SYS_SENDTO: | |
211b634b DB |
2799 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2800 | (struct sockaddr __user *)a[4], a[5]); | |
89bddce5 SH |
2801 | break; |
2802 | case SYS_RECV: | |
d27e9afc DB |
2803 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2804 | NULL, NULL); | |
89bddce5 SH |
2805 | break; |
2806 | case SYS_RECVFROM: | |
7a09e1eb DB |
2807 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2808 | (struct sockaddr __user *)a[4], | |
2809 | (int __user *)a[5]); | |
89bddce5 SH |
2810 | break; |
2811 | case SYS_SHUTDOWN: | |
005a1aea | 2812 | err = __sys_shutdown(a0, a1); |
89bddce5 SH |
2813 | break; |
2814 | case SYS_SETSOCKOPT: | |
cc36dca0 DB |
2815 | err = __sys_setsockopt(a0, a1, a[2], (char __user *)a[3], |
2816 | a[4]); | |
89bddce5 SH |
2817 | break; |
2818 | case SYS_GETSOCKOPT: | |
2819 | err = | |
13a2d70e DB |
2820 | __sys_getsockopt(a0, a1, a[2], (char __user *)a[3], |
2821 | (int __user *)a[4]); | |
89bddce5 SH |
2822 | break; |
2823 | case SYS_SENDMSG: | |
e1834a32 DB |
2824 | err = __sys_sendmsg(a0, (struct user_msghdr __user *)a1, |
2825 | a[2], true); | |
89bddce5 | 2826 | break; |
228e548e | 2827 | case SYS_SENDMMSG: |
e1834a32 DB |
2828 | err = __sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
2829 | a[3], true); | |
228e548e | 2830 | break; |
89bddce5 | 2831 | case SYS_RECVMSG: |
e1834a32 DB |
2832 | err = __sys_recvmsg(a0, (struct user_msghdr __user *)a1, |
2833 | a[2], true); | |
89bddce5 | 2834 | break; |
a2e27255 | 2835 | case SYS_RECVMMSG: |
e11d4284 AB |
2836 | if (IS_ENABLED(CONFIG_64BIT) || !IS_ENABLED(CONFIG_64BIT_TIME)) |
2837 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, | |
2838 | a[2], a[3], | |
2839 | (struct __kernel_timespec __user *)a[4], | |
2840 | NULL); | |
2841 | else | |
2842 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, | |
2843 | a[2], a[3], NULL, | |
2844 | (struct old_timespec32 __user *)a[4]); | |
a2e27255 | 2845 | break; |
de11defe | 2846 | case SYS_ACCEPT4: |
4541e805 DB |
2847 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2848 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2849 | break; |
89bddce5 SH |
2850 | default: |
2851 | err = -EINVAL; | |
2852 | break; | |
1da177e4 LT |
2853 | } |
2854 | return err; | |
2855 | } | |
2856 | ||
89bddce5 | 2857 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2858 | |
55737fda SH |
2859 | /** |
2860 | * sock_register - add a socket protocol handler | |
2861 | * @ops: description of protocol | |
2862 | * | |
1da177e4 LT |
2863 | * This function is called by a protocol handler that wants to |
2864 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2865 | * socket interface. The value ops->family corresponds to the |
55737fda | 2866 | * socket system call protocol family. |
1da177e4 | 2867 | */ |
f0fd27d4 | 2868 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2869 | { |
2870 | int err; | |
2871 | ||
2872 | if (ops->family >= NPROTO) { | |
3410f22e | 2873 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2874 | return -ENOBUFS; |
2875 | } | |
55737fda SH |
2876 | |
2877 | spin_lock(&net_family_lock); | |
190683a9 ED |
2878 | if (rcu_dereference_protected(net_families[ops->family], |
2879 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2880 | err = -EEXIST; |
2881 | else { | |
cf778b00 | 2882 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2883 | err = 0; |
2884 | } | |
55737fda SH |
2885 | spin_unlock(&net_family_lock); |
2886 | ||
3410f22e | 2887 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2888 | return err; |
2889 | } | |
c6d409cf | 2890 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2891 | |
55737fda SH |
2892 | /** |
2893 | * sock_unregister - remove a protocol handler | |
2894 | * @family: protocol family to remove | |
2895 | * | |
1da177e4 LT |
2896 | * This function is called by a protocol handler that wants to |
2897 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2898 | * new socket creation. |
2899 | * | |
2900 | * If protocol handler is a module, then it can use module reference | |
2901 | * counts to protect against new references. If protocol handler is not | |
2902 | * a module then it needs to provide its own protection in | |
2903 | * the ops->create routine. | |
1da177e4 | 2904 | */ |
f0fd27d4 | 2905 | void sock_unregister(int family) |
1da177e4 | 2906 | { |
f0fd27d4 | 2907 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2908 | |
55737fda | 2909 | spin_lock(&net_family_lock); |
a9b3cd7f | 2910 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2911 | spin_unlock(&net_family_lock); |
2912 | ||
2913 | synchronize_rcu(); | |
2914 | ||
3410f22e | 2915 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2916 | } |
c6d409cf | 2917 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2918 | |
bf2ae2e4 XL |
2919 | bool sock_is_registered(int family) |
2920 | { | |
66b51b0a | 2921 | return family < NPROTO && rcu_access_pointer(net_families[family]); |
bf2ae2e4 XL |
2922 | } |
2923 | ||
77d76ea3 | 2924 | static int __init sock_init(void) |
1da177e4 | 2925 | { |
b3e19d92 | 2926 | int err; |
2ca794e5 EB |
2927 | /* |
2928 | * Initialize the network sysctl infrastructure. | |
2929 | */ | |
2930 | err = net_sysctl_init(); | |
2931 | if (err) | |
2932 | goto out; | |
b3e19d92 | 2933 | |
1da177e4 | 2934 | /* |
89bddce5 | 2935 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2936 | */ |
2937 | skb_init(); | |
1da177e4 LT |
2938 | |
2939 | /* | |
89bddce5 | 2940 | * Initialize the protocols module. |
1da177e4 LT |
2941 | */ |
2942 | ||
2943 | init_inodecache(); | |
b3e19d92 NP |
2944 | |
2945 | err = register_filesystem(&sock_fs_type); | |
2946 | if (err) | |
2947 | goto out_fs; | |
1da177e4 | 2948 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2949 | if (IS_ERR(sock_mnt)) { |
2950 | err = PTR_ERR(sock_mnt); | |
2951 | goto out_mount; | |
2952 | } | |
77d76ea3 AK |
2953 | |
2954 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2955 | */ |
2956 | ||
2957 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
2958 | err = netfilter_init(); |
2959 | if (err) | |
2960 | goto out; | |
1da177e4 | 2961 | #endif |
cbeb321a | 2962 | |
408eccce | 2963 | ptp_classifier_init(); |
c1f19b51 | 2964 | |
b3e19d92 NP |
2965 | out: |
2966 | return err; | |
2967 | ||
2968 | out_mount: | |
2969 | unregister_filesystem(&sock_fs_type); | |
2970 | out_fs: | |
2971 | goto out; | |
1da177e4 LT |
2972 | } |
2973 | ||
77d76ea3 AK |
2974 | core_initcall(sock_init); /* early initcall */ |
2975 | ||
1da177e4 LT |
2976 | #ifdef CONFIG_PROC_FS |
2977 | void socket_seq_show(struct seq_file *seq) | |
2978 | { | |
648845ab TZ |
2979 | seq_printf(seq, "sockets: used %d\n", |
2980 | sock_inuse_get(seq->private)); | |
1da177e4 | 2981 | } |
89bddce5 | 2982 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 2983 | |
89bbfc95 | 2984 | #ifdef CONFIG_COMPAT |
36fd633e | 2985 | static int compat_dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 2986 | { |
6b96018b | 2987 | struct compat_ifconf ifc32; |
7a229387 | 2988 | struct ifconf ifc; |
7a229387 AB |
2989 | int err; |
2990 | ||
6b96018b | 2991 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2992 | return -EFAULT; |
2993 | ||
36fd633e AV |
2994 | ifc.ifc_len = ifc32.ifc_len; |
2995 | ifc.ifc_req = compat_ptr(ifc32.ifcbuf); | |
7a229387 | 2996 | |
36fd633e AV |
2997 | rtnl_lock(); |
2998 | err = dev_ifconf(net, &ifc, sizeof(struct compat_ifreq)); | |
2999 | rtnl_unlock(); | |
7a229387 AB |
3000 | if (err) |
3001 | return err; | |
3002 | ||
36fd633e | 3003 | ifc32.ifc_len = ifc.ifc_len; |
6b96018b | 3004 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
3005 | return -EFAULT; |
3006 | ||
3007 | return 0; | |
3008 | } | |
3009 | ||
6b96018b | 3010 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 3011 | { |
3a7da39d BH |
3012 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
3013 | bool convert_in = false, convert_out = false; | |
44c02a2c AV |
3014 | size_t buf_size = 0; |
3015 | struct ethtool_rxnfc __user *rxnfc = NULL; | |
3016 | struct ifreq ifr; | |
3a7da39d BH |
3017 | u32 rule_cnt = 0, actual_rule_cnt; |
3018 | u32 ethcmd; | |
7a229387 | 3019 | u32 data; |
3a7da39d | 3020 | int ret; |
7a229387 | 3021 | |
3a7da39d BH |
3022 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
3023 | return -EFAULT; | |
7a229387 | 3024 | |
3a7da39d BH |
3025 | compat_rxnfc = compat_ptr(data); |
3026 | ||
3027 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
3028 | return -EFAULT; |
3029 | ||
3a7da39d BH |
3030 | /* Most ethtool structures are defined without padding. |
3031 | * Unfortunately struct ethtool_rxnfc is an exception. | |
3032 | */ | |
3033 | switch (ethcmd) { | |
3034 | default: | |
3035 | break; | |
3036 | case ETHTOOL_GRXCLSRLALL: | |
3037 | /* Buffer size is variable */ | |
3038 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
3039 | return -EFAULT; | |
3040 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
3041 | return -ENOMEM; | |
3042 | buf_size += rule_cnt * sizeof(u32); | |
3043 | /* fall through */ | |
3044 | case ETHTOOL_GRXRINGS: | |
3045 | case ETHTOOL_GRXCLSRLCNT: | |
3046 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 3047 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
3048 | convert_out = true; |
3049 | /* fall through */ | |
3050 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
3051 | buf_size += sizeof(struct ethtool_rxnfc); |
3052 | convert_in = true; | |
44c02a2c | 3053 | rxnfc = compat_alloc_user_space(buf_size); |
3a7da39d BH |
3054 | break; |
3055 | } | |
3056 | ||
44c02a2c | 3057 | if (copy_from_user(&ifr.ifr_name, &ifr32->ifr_name, IFNAMSIZ)) |
7a229387 AB |
3058 | return -EFAULT; |
3059 | ||
44c02a2c | 3060 | ifr.ifr_data = convert_in ? rxnfc : (void __user *)compat_rxnfc; |
7a229387 | 3061 | |
3a7da39d | 3062 | if (convert_in) { |
127fe533 | 3063 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
3064 | * fs.ring_cookie and at the end of fs, but nowhere else. |
3065 | */ | |
127fe533 AD |
3066 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
3067 | sizeof(compat_rxnfc->fs.m_ext) != | |
3068 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
3069 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
3070 | BUILD_BUG_ON( |
3071 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
3072 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
3073 | offsetof(struct ethtool_rxnfc, fs.location) - | |
3074 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
3075 | ||
3076 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
3077 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
3078 | (void __user *)rxnfc) || | |
3a7da39d BH |
3079 | copy_in_user(&rxnfc->fs.ring_cookie, |
3080 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 | 3081 | (void __user *)(&rxnfc->fs.location + 1) - |
b6168562 WW |
3082 | (void __user *)&rxnfc->fs.ring_cookie)) |
3083 | return -EFAULT; | |
3084 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3085 | if (put_user(rule_cnt, &rxnfc->rule_cnt)) | |
3086 | return -EFAULT; | |
3087 | } else if (copy_in_user(&rxnfc->rule_cnt, | |
3088 | &compat_rxnfc->rule_cnt, | |
3089 | sizeof(rxnfc->rule_cnt))) | |
3a7da39d BH |
3090 | return -EFAULT; |
3091 | } | |
3092 | ||
44c02a2c | 3093 | ret = dev_ioctl(net, SIOCETHTOOL, &ifr, NULL); |
3a7da39d BH |
3094 | if (ret) |
3095 | return ret; | |
3096 | ||
3097 | if (convert_out) { | |
3098 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
3099 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
3100 | (const void __user *)rxnfc) || | |
3a7da39d BH |
3101 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
3102 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
3103 | (const void __user *)(&rxnfc->fs.location + 1) - |
3104 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
3105 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
3106 | sizeof(rxnfc->rule_cnt))) | |
3107 | return -EFAULT; | |
3108 | ||
3109 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3110 | /* As an optimisation, we only copy the actual | |
3111 | * number of rules that the underlying | |
3112 | * function returned. Since Mallory might | |
3113 | * change the rule count in user memory, we | |
3114 | * check that it is less than the rule count | |
3115 | * originally given (as the user buffer size), | |
3116 | * which has been range-checked. | |
3117 | */ | |
3118 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
3119 | return -EFAULT; | |
3120 | if (actual_rule_cnt < rule_cnt) | |
3121 | rule_cnt = actual_rule_cnt; | |
3122 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
3123 | &rxnfc->rule_locs[0], | |
3124 | rule_cnt * sizeof(u32))) | |
3125 | return -EFAULT; | |
3126 | } | |
3127 | } | |
3128 | ||
3129 | return 0; | |
7a229387 AB |
3130 | } |
3131 | ||
7a50a240 AB |
3132 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
3133 | { | |
7a50a240 | 3134 | compat_uptr_t uptr32; |
44c02a2c AV |
3135 | struct ifreq ifr; |
3136 | void __user *saved; | |
3137 | int err; | |
7a50a240 | 3138 | |
44c02a2c | 3139 | if (copy_from_user(&ifr, uifr32, sizeof(struct compat_ifreq))) |
7a50a240 AB |
3140 | return -EFAULT; |
3141 | ||
3142 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
3143 | return -EFAULT; | |
3144 | ||
44c02a2c AV |
3145 | saved = ifr.ifr_settings.ifs_ifsu.raw_hdlc; |
3146 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = compat_ptr(uptr32); | |
7a229387 | 3147 | |
44c02a2c AV |
3148 | err = dev_ioctl(net, SIOCWANDEV, &ifr, NULL); |
3149 | if (!err) { | |
3150 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = saved; | |
3151 | if (copy_to_user(uifr32, &ifr, sizeof(struct compat_ifreq))) | |
3152 | err = -EFAULT; | |
ccbd6a5a | 3153 | } |
44c02a2c | 3154 | return err; |
7a229387 AB |
3155 | } |
3156 | ||
590d4693 BH |
3157 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
3158 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 3159 | struct compat_ifreq __user *u_ifreq32) |
7a229387 | 3160 | { |
44c02a2c | 3161 | struct ifreq ifreq; |
7a229387 AB |
3162 | u32 data32; |
3163 | ||
44c02a2c | 3164 | if (copy_from_user(ifreq.ifr_name, u_ifreq32->ifr_name, IFNAMSIZ)) |
7a229387 | 3165 | return -EFAULT; |
44c02a2c | 3166 | if (get_user(data32, &u_ifreq32->ifr_data)) |
7a229387 | 3167 | return -EFAULT; |
44c02a2c | 3168 | ifreq.ifr_data = compat_ptr(data32); |
7a229387 | 3169 | |
44c02a2c | 3170 | return dev_ioctl(net, cmd, &ifreq, NULL); |
7a229387 AB |
3171 | } |
3172 | ||
37ac39bd JB |
3173 | static int compat_ifreq_ioctl(struct net *net, struct socket *sock, |
3174 | unsigned int cmd, | |
3175 | struct compat_ifreq __user *uifr32) | |
3176 | { | |
3177 | struct ifreq __user *uifr; | |
3178 | int err; | |
3179 | ||
3180 | /* Handle the fact that while struct ifreq has the same *layout* on | |
3181 | * 32/64 for everything but ifreq::ifru_ifmap and ifreq::ifru_data, | |
3182 | * which are handled elsewhere, it still has different *size* due to | |
3183 | * ifreq::ifru_ifmap (which is 16 bytes on 32 bit, 24 bytes on 64-bit, | |
3184 | * resulting in struct ifreq being 32 and 40 bytes respectively). | |
3185 | * As a result, if the struct happens to be at the end of a page and | |
3186 | * the next page isn't readable/writable, we get a fault. To prevent | |
3187 | * that, copy back and forth to the full size. | |
3188 | */ | |
3189 | ||
3190 | uifr = compat_alloc_user_space(sizeof(*uifr)); | |
3191 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
3192 | return -EFAULT; | |
3193 | ||
3194 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
3195 | ||
3196 | if (!err) { | |
3197 | switch (cmd) { | |
3198 | case SIOCGIFFLAGS: | |
3199 | case SIOCGIFMETRIC: | |
3200 | case SIOCGIFMTU: | |
3201 | case SIOCGIFMEM: | |
3202 | case SIOCGIFHWADDR: | |
3203 | case SIOCGIFINDEX: | |
3204 | case SIOCGIFADDR: | |
3205 | case SIOCGIFBRDADDR: | |
3206 | case SIOCGIFDSTADDR: | |
3207 | case SIOCGIFNETMASK: | |
3208 | case SIOCGIFPFLAGS: | |
3209 | case SIOCGIFTXQLEN: | |
3210 | case SIOCGMIIPHY: | |
3211 | case SIOCGMIIREG: | |
c6c9fee3 | 3212 | case SIOCGIFNAME: |
37ac39bd JB |
3213 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
3214 | err = -EFAULT; | |
3215 | break; | |
3216 | } | |
3217 | } | |
3218 | return err; | |
3219 | } | |
3220 | ||
a2116ed2 AB |
3221 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
3222 | struct compat_ifreq __user *uifr32) | |
3223 | { | |
3224 | struct ifreq ifr; | |
3225 | struct compat_ifmap __user *uifmap32; | |
a2116ed2 AB |
3226 | int err; |
3227 | ||
3228 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
3229 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3230 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3231 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3232 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3233 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3234 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3235 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3236 | if (err) |
3237 | return -EFAULT; | |
3238 | ||
44c02a2c | 3239 | err = dev_ioctl(net, cmd, &ifr, NULL); |
a2116ed2 AB |
3240 | |
3241 | if (cmd == SIOCGIFMAP && !err) { | |
3242 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3243 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3244 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3245 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3246 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3247 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3248 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3249 | if (err) |
3250 | err = -EFAULT; | |
3251 | } | |
3252 | return err; | |
3253 | } | |
3254 | ||
7a229387 | 3255 | struct rtentry32 { |
c6d409cf | 3256 | u32 rt_pad1; |
7a229387 AB |
3257 | struct sockaddr rt_dst; /* target address */ |
3258 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
3259 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
3260 | unsigned short rt_flags; |
3261 | short rt_pad2; | |
3262 | u32 rt_pad3; | |
3263 | unsigned char rt_tos; | |
3264 | unsigned char rt_class; | |
3265 | short rt_pad4; | |
3266 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 3267 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
3268 | u32 rt_mtu; /* per route MTU/Window */ |
3269 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
3270 | unsigned short rt_irtt; /* Initial RTT */ |
3271 | }; | |
3272 | ||
3273 | struct in6_rtmsg32 { | |
3274 | struct in6_addr rtmsg_dst; | |
3275 | struct in6_addr rtmsg_src; | |
3276 | struct in6_addr rtmsg_gateway; | |
3277 | u32 rtmsg_type; | |
3278 | u16 rtmsg_dst_len; | |
3279 | u16 rtmsg_src_len; | |
3280 | u32 rtmsg_metric; | |
3281 | u32 rtmsg_info; | |
3282 | u32 rtmsg_flags; | |
3283 | s32 rtmsg_ifindex; | |
3284 | }; | |
3285 | ||
6b96018b AB |
3286 | static int routing_ioctl(struct net *net, struct socket *sock, |
3287 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3288 | { |
3289 | int ret; | |
3290 | void *r = NULL; | |
3291 | struct in6_rtmsg r6; | |
3292 | struct rtentry r4; | |
3293 | char devname[16]; | |
3294 | u32 rtdev; | |
3295 | mm_segment_t old_fs = get_fs(); | |
3296 | ||
6b96018b AB |
3297 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3298 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3299 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3300 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
3301 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3302 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3303 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3304 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3305 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3306 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3307 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3308 | |
3309 | r = (void *) &r6; | |
3310 | } else { /* ipv4 */ | |
6b96018b | 3311 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3312 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3313 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
3314 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
3315 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3316 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3317 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
3318 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3319 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3320 | if (rtdev) { |
c6d409cf | 3321 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3322 | r4.rt_dev = (char __user __force *)devname; |
3323 | devname[15] = 0; | |
7a229387 AB |
3324 | } else |
3325 | r4.rt_dev = NULL; | |
3326 | ||
3327 | r = (void *) &r4; | |
3328 | } | |
3329 | ||
3330 | if (ret) { | |
3331 | ret = -EFAULT; | |
3332 | goto out; | |
3333 | } | |
3334 | ||
c6d409cf | 3335 | set_fs(KERNEL_DS); |
63ff03ab | 3336 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3337 | set_fs(old_fs); |
7a229387 AB |
3338 | |
3339 | out: | |
7a229387 AB |
3340 | return ret; |
3341 | } | |
3342 | ||
3343 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3344 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3345 | * use compatible ioctls |
7a229387 | 3346 | */ |
6b96018b | 3347 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3348 | { |
6b96018b | 3349 | compat_ulong_t tmp; |
7a229387 | 3350 | |
6b96018b | 3351 | if (get_user(tmp, argp)) |
7a229387 AB |
3352 | return -EFAULT; |
3353 | if (tmp == BRCTL_GET_VERSION) | |
3354 | return BRCTL_VERSION + 1; | |
3355 | return -EINVAL; | |
3356 | } | |
3357 | ||
6b96018b AB |
3358 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3359 | unsigned int cmd, unsigned long arg) | |
3360 | { | |
3361 | void __user *argp = compat_ptr(arg); | |
3362 | struct sock *sk = sock->sk; | |
3363 | struct net *net = sock_net(sk); | |
7a229387 | 3364 | |
6b96018b | 3365 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3366 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3367 | |
3368 | switch (cmd) { | |
3369 | case SIOCSIFBR: | |
3370 | case SIOCGIFBR: | |
3371 | return old_bridge_ioctl(argp); | |
6b96018b | 3372 | case SIOCGIFCONF: |
36fd633e | 3373 | return compat_dev_ifconf(net, argp); |
6b96018b AB |
3374 | case SIOCETHTOOL: |
3375 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3376 | case SIOCWANDEV: |
3377 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3378 | case SIOCGIFMAP: |
3379 | case SIOCSIFMAP: | |
3380 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3381 | case SIOCADDRT: |
3382 | case SIOCDELRT: | |
3383 | return routing_ioctl(net, sock, cmd, argp); | |
0768e170 AB |
3384 | case SIOCGSTAMP_OLD: |
3385 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
3386 | if (!sock->ops->gettstamp) |
3387 | return -ENOIOCTLCMD; | |
0768e170 | 3388 | return sock->ops->gettstamp(sock, argp, cmd == SIOCGSTAMP_OLD, |
c7cbdbf2 AB |
3389 | !COMPAT_USE_64BIT_TIME); |
3390 | ||
590d4693 BH |
3391 | case SIOCBONDSLAVEINFOQUERY: |
3392 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3393 | case SIOCSHWTSTAMP: |
fd468c74 | 3394 | case SIOCGHWTSTAMP: |
590d4693 | 3395 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3396 | |
3397 | case FIOSETOWN: | |
3398 | case SIOCSPGRP: | |
3399 | case FIOGETOWN: | |
3400 | case SIOCGPGRP: | |
3401 | case SIOCBRADDBR: | |
3402 | case SIOCBRDELBR: | |
3403 | case SIOCGIFVLAN: | |
3404 | case SIOCSIFVLAN: | |
3405 | case SIOCADDDLCI: | |
3406 | case SIOCDELDLCI: | |
c62cce2c | 3407 | case SIOCGSKNS: |
0768e170 AB |
3408 | case SIOCGSTAMP_NEW: |
3409 | case SIOCGSTAMPNS_NEW: | |
6b96018b AB |
3410 | return sock_ioctl(file, cmd, arg); |
3411 | ||
3412 | case SIOCGIFFLAGS: | |
3413 | case SIOCSIFFLAGS: | |
3414 | case SIOCGIFMETRIC: | |
3415 | case SIOCSIFMETRIC: | |
3416 | case SIOCGIFMTU: | |
3417 | case SIOCSIFMTU: | |
3418 | case SIOCGIFMEM: | |
3419 | case SIOCSIFMEM: | |
3420 | case SIOCGIFHWADDR: | |
3421 | case SIOCSIFHWADDR: | |
3422 | case SIOCADDMULTI: | |
3423 | case SIOCDELMULTI: | |
3424 | case SIOCGIFINDEX: | |
6b96018b AB |
3425 | case SIOCGIFADDR: |
3426 | case SIOCSIFADDR: | |
3427 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3428 | case SIOCDIFADDR: |
6b96018b AB |
3429 | case SIOCGIFBRDADDR: |
3430 | case SIOCSIFBRDADDR: | |
3431 | case SIOCGIFDSTADDR: | |
3432 | case SIOCSIFDSTADDR: | |
3433 | case SIOCGIFNETMASK: | |
3434 | case SIOCSIFNETMASK: | |
3435 | case SIOCSIFPFLAGS: | |
3436 | case SIOCGIFPFLAGS: | |
3437 | case SIOCGIFTXQLEN: | |
3438 | case SIOCSIFTXQLEN: | |
3439 | case SIOCBRADDIF: | |
3440 | case SIOCBRDELIF: | |
c6c9fee3 | 3441 | case SIOCGIFNAME: |
9177efd3 AB |
3442 | case SIOCSIFNAME: |
3443 | case SIOCGMIIPHY: | |
3444 | case SIOCGMIIREG: | |
3445 | case SIOCSMIIREG: | |
f92d4fc9 AV |
3446 | case SIOCBONDENSLAVE: |
3447 | case SIOCBONDRELEASE: | |
3448 | case SIOCBONDSETHWADDR: | |
3449 | case SIOCBONDCHANGEACTIVE: | |
37ac39bd JB |
3450 | return compat_ifreq_ioctl(net, sock, cmd, argp); |
3451 | ||
6b96018b AB |
3452 | case SIOCSARP: |
3453 | case SIOCGARP: | |
3454 | case SIOCDARP: | |
6b96018b | 3455 | case SIOCATMARK: |
63ff03ab | 3456 | return sock_do_ioctl(net, sock, cmd, arg); |
9177efd3 AB |
3457 | } |
3458 | ||
6b96018b AB |
3459 | return -ENOIOCTLCMD; |
3460 | } | |
7a229387 | 3461 | |
95c96174 | 3462 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3463 | unsigned long arg) |
89bbfc95 SP |
3464 | { |
3465 | struct socket *sock = file->private_data; | |
3466 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3467 | struct sock *sk; |
3468 | struct net *net; | |
3469 | ||
3470 | sk = sock->sk; | |
3471 | net = sock_net(sk); | |
89bbfc95 SP |
3472 | |
3473 | if (sock->ops->compat_ioctl) | |
3474 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3475 | ||
87de87d5 DM |
3476 | if (ret == -ENOIOCTLCMD && |
3477 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3478 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3479 | ||
6b96018b AB |
3480 | if (ret == -ENOIOCTLCMD) |
3481 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3482 | ||
89bbfc95 SP |
3483 | return ret; |
3484 | } | |
3485 | #endif | |
3486 | ||
8a3c245c PT |
3487 | /** |
3488 | * kernel_bind - bind an address to a socket (kernel space) | |
3489 | * @sock: socket | |
3490 | * @addr: address | |
3491 | * @addrlen: length of address | |
3492 | * | |
3493 | * Returns 0 or an error. | |
3494 | */ | |
3495 | ||
ac5a488e SS |
3496 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3497 | { | |
3498 | return sock->ops->bind(sock, addr, addrlen); | |
3499 | } | |
c6d409cf | 3500 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e | 3501 | |
8a3c245c PT |
3502 | /** |
3503 | * kernel_listen - move socket to listening state (kernel space) | |
3504 | * @sock: socket | |
3505 | * @backlog: pending connections queue size | |
3506 | * | |
3507 | * Returns 0 or an error. | |
3508 | */ | |
3509 | ||
ac5a488e SS |
3510 | int kernel_listen(struct socket *sock, int backlog) |
3511 | { | |
3512 | return sock->ops->listen(sock, backlog); | |
3513 | } | |
c6d409cf | 3514 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e | 3515 | |
8a3c245c PT |
3516 | /** |
3517 | * kernel_accept - accept a connection (kernel space) | |
3518 | * @sock: listening socket | |
3519 | * @newsock: new connected socket | |
3520 | * @flags: flags | |
3521 | * | |
3522 | * @flags must be SOCK_CLOEXEC, SOCK_NONBLOCK or 0. | |
3523 | * If it fails, @newsock is guaranteed to be %NULL. | |
3524 | * Returns 0 or an error. | |
3525 | */ | |
3526 | ||
ac5a488e SS |
3527 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) |
3528 | { | |
3529 | struct sock *sk = sock->sk; | |
3530 | int err; | |
3531 | ||
3532 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3533 | newsock); | |
3534 | if (err < 0) | |
3535 | goto done; | |
3536 | ||
cdfbabfb | 3537 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3538 | if (err < 0) { |
3539 | sock_release(*newsock); | |
fa8705b0 | 3540 | *newsock = NULL; |
ac5a488e SS |
3541 | goto done; |
3542 | } | |
3543 | ||
3544 | (*newsock)->ops = sock->ops; | |
1b08534e | 3545 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3546 | |
3547 | done: | |
3548 | return err; | |
3549 | } | |
c6d409cf | 3550 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e | 3551 | |
8a3c245c PT |
3552 | /** |
3553 | * kernel_connect - connect a socket (kernel space) | |
3554 | * @sock: socket | |
3555 | * @addr: address | |
3556 | * @addrlen: address length | |
3557 | * @flags: flags (O_NONBLOCK, ...) | |
3558 | * | |
3559 | * For datagram sockets, @addr is the addres to which datagrams are sent | |
3560 | * by default, and the only address from which datagrams are received. | |
3561 | * For stream sockets, attempts to connect to @addr. | |
3562 | * Returns 0 or an error code. | |
3563 | */ | |
3564 | ||
ac5a488e | 3565 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, |
4768fbcb | 3566 | int flags) |
ac5a488e SS |
3567 | { |
3568 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3569 | } | |
c6d409cf | 3570 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e | 3571 | |
8a3c245c PT |
3572 | /** |
3573 | * kernel_getsockname - get the address which the socket is bound (kernel space) | |
3574 | * @sock: socket | |
3575 | * @addr: address holder | |
3576 | * | |
3577 | * Fills the @addr pointer with the address which the socket is bound. | |
3578 | * Returns 0 or an error code. | |
3579 | */ | |
3580 | ||
9b2c45d4 | 3581 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3582 | { |
9b2c45d4 | 3583 | return sock->ops->getname(sock, addr, 0); |
ac5a488e | 3584 | } |
c6d409cf | 3585 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e | 3586 | |
8a3c245c PT |
3587 | /** |
3588 | * kernel_peername - get the address which the socket is connected (kernel space) | |
3589 | * @sock: socket | |
3590 | * @addr: address holder | |
3591 | * | |
3592 | * Fills the @addr pointer with the address which the socket is connected. | |
3593 | * Returns 0 or an error code. | |
3594 | */ | |
3595 | ||
9b2c45d4 | 3596 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3597 | { |
9b2c45d4 | 3598 | return sock->ops->getname(sock, addr, 1); |
ac5a488e | 3599 | } |
c6d409cf | 3600 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e | 3601 | |
8a3c245c PT |
3602 | /** |
3603 | * kernel_getsockopt - get a socket option (kernel space) | |
3604 | * @sock: socket | |
3605 | * @level: API level (SOL_SOCKET, ...) | |
3606 | * @optname: option tag | |
3607 | * @optval: option value | |
3608 | * @optlen: option length | |
3609 | * | |
3610 | * Assigns the option length to @optlen. | |
3611 | * Returns 0 or an error. | |
3612 | */ | |
3613 | ||
ac5a488e SS |
3614 | int kernel_getsockopt(struct socket *sock, int level, int optname, |
3615 | char *optval, int *optlen) | |
3616 | { | |
3617 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3618 | char __user *uoptval; |
3619 | int __user *uoptlen; | |
ac5a488e SS |
3620 | int err; |
3621 | ||
fb8621bb NK |
3622 | uoptval = (char __user __force *) optval; |
3623 | uoptlen = (int __user __force *) optlen; | |
3624 | ||
ac5a488e SS |
3625 | set_fs(KERNEL_DS); |
3626 | if (level == SOL_SOCKET) | |
fb8621bb | 3627 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3628 | else |
fb8621bb NK |
3629 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3630 | uoptlen); | |
ac5a488e SS |
3631 | set_fs(oldfs); |
3632 | return err; | |
3633 | } | |
c6d409cf | 3634 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e | 3635 | |
8a3c245c PT |
3636 | /** |
3637 | * kernel_setsockopt - set a socket option (kernel space) | |
3638 | * @sock: socket | |
3639 | * @level: API level (SOL_SOCKET, ...) | |
3640 | * @optname: option tag | |
3641 | * @optval: option value | |
3642 | * @optlen: option length | |
3643 | * | |
3644 | * Returns 0 or an error. | |
3645 | */ | |
3646 | ||
ac5a488e | 3647 | int kernel_setsockopt(struct socket *sock, int level, int optname, |
b7058842 | 3648 | char *optval, unsigned int optlen) |
ac5a488e SS |
3649 | { |
3650 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3651 | char __user *uoptval; |
ac5a488e SS |
3652 | int err; |
3653 | ||
fb8621bb NK |
3654 | uoptval = (char __user __force *) optval; |
3655 | ||
ac5a488e SS |
3656 | set_fs(KERNEL_DS); |
3657 | if (level == SOL_SOCKET) | |
fb8621bb | 3658 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3659 | else |
fb8621bb | 3660 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3661 | optlen); |
3662 | set_fs(oldfs); | |
3663 | return err; | |
3664 | } | |
c6d409cf | 3665 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e | 3666 | |
8a3c245c PT |
3667 | /** |
3668 | * kernel_sendpage - send a &page through a socket (kernel space) | |
3669 | * @sock: socket | |
3670 | * @page: page | |
3671 | * @offset: page offset | |
3672 | * @size: total size in bytes | |
3673 | * @flags: flags (MSG_DONTWAIT, ...) | |
3674 | * | |
3675 | * Returns the total amount sent in bytes or an error. | |
3676 | */ | |
3677 | ||
ac5a488e SS |
3678 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, |
3679 | size_t size, int flags) | |
3680 | { | |
3681 | if (sock->ops->sendpage) | |
3682 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3683 | ||
3684 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3685 | } | |
c6d409cf | 3686 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3687 | |
8a3c245c PT |
3688 | /** |
3689 | * kernel_sendpage_locked - send a &page through the locked sock (kernel space) | |
3690 | * @sk: sock | |
3691 | * @page: page | |
3692 | * @offset: page offset | |
3693 | * @size: total size in bytes | |
3694 | * @flags: flags (MSG_DONTWAIT, ...) | |
3695 | * | |
3696 | * Returns the total amount sent in bytes or an error. | |
3697 | * Caller must hold @sk. | |
3698 | */ | |
3699 | ||
306b13eb TH |
3700 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3701 | size_t size, int flags) | |
3702 | { | |
3703 | struct socket *sock = sk->sk_socket; | |
3704 | ||
3705 | if (sock->ops->sendpage_locked) | |
3706 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3707 | flags); | |
3708 | ||
3709 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3710 | } | |
3711 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3712 | ||
8a3c245c PT |
3713 | /** |
3714 | * kernel_shutdown - shut down part of a full-duplex connection (kernel space) | |
3715 | * @sock: socket | |
3716 | * @how: connection part | |
3717 | * | |
3718 | * Returns 0 or an error. | |
3719 | */ | |
3720 | ||
91cf45f0 TM |
3721 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3722 | { | |
3723 | return sock->ops->shutdown(sock, how); | |
3724 | } | |
91cf45f0 | 3725 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 | 3726 | |
8a3c245c PT |
3727 | /** |
3728 | * kernel_sock_ip_overhead - returns the IP overhead imposed by a socket | |
3729 | * @sk: socket | |
3730 | * | |
3731 | * This routine returns the IP overhead imposed by a socket i.e. | |
3732 | * the length of the underlying IP header, depending on whether | |
3733 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
3734 | * on at the socket. Assumes that the caller has a lock on the socket. | |
113c3075 | 3735 | */ |
8a3c245c | 3736 | |
113c3075 P |
3737 | u32 kernel_sock_ip_overhead(struct sock *sk) |
3738 | { | |
3739 | struct inet_sock *inet; | |
3740 | struct ip_options_rcu *opt; | |
3741 | u32 overhead = 0; | |
113c3075 P |
3742 | #if IS_ENABLED(CONFIG_IPV6) |
3743 | struct ipv6_pinfo *np; | |
3744 | struct ipv6_txoptions *optv6 = NULL; | |
3745 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3746 | ||
3747 | if (!sk) | |
3748 | return overhead; | |
3749 | ||
113c3075 P |
3750 | switch (sk->sk_family) { |
3751 | case AF_INET: | |
3752 | inet = inet_sk(sk); | |
3753 | overhead += sizeof(struct iphdr); | |
3754 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3755 | sock_owned_by_user(sk)); |
113c3075 P |
3756 | if (opt) |
3757 | overhead += opt->opt.optlen; | |
3758 | return overhead; | |
3759 | #if IS_ENABLED(CONFIG_IPV6) | |
3760 | case AF_INET6: | |
3761 | np = inet6_sk(sk); | |
3762 | overhead += sizeof(struct ipv6hdr); | |
3763 | if (np) | |
3764 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3765 | sock_owned_by_user(sk)); |
113c3075 P |
3766 | if (optv6) |
3767 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3768 | return overhead; | |
3769 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3770 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3771 | return overhead; | |
3772 | } | |
3773 | } | |
3774 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |