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> | |
c7dc504e | 103 | #include <linux/termios.h> |
6b96018b | 104 | #include <linux/sockios.h> |
076bb0c8 | 105 | #include <net/busy_poll.h> |
f24b9be5 | 106 | #include <linux/errqueue.h> |
06021292 | 107 | |
e0d1095a | 108 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
109 | unsigned int sysctl_net_busy_read __read_mostly; |
110 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 111 | #endif |
6b96018b | 112 | |
8ae5e030 AV |
113 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
114 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 115 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
116 | |
117 | static int sock_close(struct inode *inode, struct file *file); | |
a11e1d43 LT |
118 | static __poll_t sock_poll(struct file *file, |
119 | struct poll_table_struct *wait); | |
89bddce5 | 120 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
121 | #ifdef CONFIG_COMPAT |
122 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 123 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 124 | #endif |
1da177e4 | 125 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
126 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
127 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 128 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 129 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 130 | unsigned int flags); |
1da177e4 | 131 | |
1da177e4 LT |
132 | /* |
133 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
134 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
135 | */ | |
136 | ||
da7071d7 | 137 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
138 | .owner = THIS_MODULE, |
139 | .llseek = no_llseek, | |
8ae5e030 AV |
140 | .read_iter = sock_read_iter, |
141 | .write_iter = sock_write_iter, | |
1da177e4 LT |
142 | .poll = sock_poll, |
143 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
144 | #ifdef CONFIG_COMPAT |
145 | .compat_ioctl = compat_sock_ioctl, | |
146 | #endif | |
1da177e4 | 147 | .mmap = sock_mmap, |
1da177e4 LT |
148 | .release = sock_close, |
149 | .fasync = sock_fasync, | |
5274f052 JA |
150 | .sendpage = sock_sendpage, |
151 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 152 | .splice_read = sock_splice_read, |
1da177e4 LT |
153 | }; |
154 | ||
155 | /* | |
156 | * The protocol list. Each protocol is registered in here. | |
157 | */ | |
158 | ||
1da177e4 | 159 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 160 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 161 | |
1da177e4 | 162 | /* |
89bddce5 SH |
163 | * Support routines. |
164 | * Move socket addresses back and forth across the kernel/user | |
165 | * divide and look after the messy bits. | |
1da177e4 LT |
166 | */ |
167 | ||
1da177e4 LT |
168 | /** |
169 | * move_addr_to_kernel - copy a socket address into kernel space | |
170 | * @uaddr: Address in user space | |
171 | * @kaddr: Address in kernel space | |
172 | * @ulen: Length in user space | |
173 | * | |
174 | * The address is copied into kernel space. If the provided address is | |
175 | * too long an error code of -EINVAL is returned. If the copy gives | |
176 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
177 | */ | |
178 | ||
43db362d | 179 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 180 | { |
230b1839 | 181 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 182 | return -EINVAL; |
89bddce5 | 183 | if (ulen == 0) |
1da177e4 | 184 | return 0; |
89bddce5 | 185 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 186 | return -EFAULT; |
3ec3b2fb | 187 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
188 | } |
189 | ||
190 | /** | |
191 | * move_addr_to_user - copy an address to user space | |
192 | * @kaddr: kernel space address | |
193 | * @klen: length of address in kernel | |
194 | * @uaddr: user space address | |
195 | * @ulen: pointer to user length field | |
196 | * | |
197 | * The value pointed to by ulen on entry is the buffer length available. | |
198 | * This is overwritten with the buffer space used. -EINVAL is returned | |
199 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
200 | * is returned if either the buffer or the length field are not | |
201 | * accessible. | |
202 | * After copying the data up to the limit the user specifies, the true | |
203 | * length of the data is written over the length limit the user | |
204 | * specified. Zero is returned for a success. | |
205 | */ | |
89bddce5 | 206 | |
43db362d | 207 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 208 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
209 | { |
210 | int err; | |
211 | int len; | |
212 | ||
68c6beb3 | 213 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
214 | err = get_user(len, ulen); |
215 | if (err) | |
1da177e4 | 216 | return err; |
89bddce5 SH |
217 | if (len > klen) |
218 | len = klen; | |
68c6beb3 | 219 | if (len < 0) |
1da177e4 | 220 | return -EINVAL; |
89bddce5 | 221 | if (len) { |
d6fe3945 SG |
222 | if (audit_sockaddr(klen, kaddr)) |
223 | return -ENOMEM; | |
89bddce5 | 224 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
225 | return -EFAULT; |
226 | } | |
227 | /* | |
89bddce5 SH |
228 | * "fromlen shall refer to the value before truncation.." |
229 | * 1003.1g | |
1da177e4 LT |
230 | */ |
231 | return __put_user(klen, ulen); | |
232 | } | |
233 | ||
08009a76 | 234 | static struct kmem_cache *sock_inode_cachep __ro_after_init; |
1da177e4 LT |
235 | |
236 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
237 | { | |
238 | struct socket_alloc *ei; | |
89bddce5 | 239 | |
e94b1766 | 240 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
241 | if (!ei) |
242 | return NULL; | |
333f7909 AV |
243 | init_waitqueue_head(&ei->socket.wq.wait); |
244 | ei->socket.wq.fasync_list = NULL; | |
245 | ei->socket.wq.flags = 0; | |
89bddce5 | 246 | |
1da177e4 LT |
247 | ei->socket.state = SS_UNCONNECTED; |
248 | ei->socket.flags = 0; | |
249 | ei->socket.ops = NULL; | |
250 | ei->socket.sk = NULL; | |
251 | ei->socket.file = NULL; | |
1da177e4 LT |
252 | |
253 | return &ei->vfs_inode; | |
254 | } | |
255 | ||
6d7855c5 | 256 | static void sock_free_inode(struct inode *inode) |
1da177e4 | 257 | { |
43815482 ED |
258 | struct socket_alloc *ei; |
259 | ||
260 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
43815482 | 261 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
262 | } |
263 | ||
51cc5068 | 264 | static void init_once(void *foo) |
1da177e4 | 265 | { |
89bddce5 | 266 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 267 | |
a35afb83 | 268 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 269 | } |
89bddce5 | 270 | |
1e911632 | 271 | static void init_inodecache(void) |
1da177e4 LT |
272 | { |
273 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
274 | sizeof(struct socket_alloc), |
275 | 0, | |
276 | (SLAB_HWCACHE_ALIGN | | |
277 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 278 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 279 | init_once); |
1e911632 | 280 | BUG_ON(sock_inode_cachep == NULL); |
1da177e4 LT |
281 | } |
282 | ||
b87221de | 283 | static const struct super_operations sockfs_ops = { |
c6d409cf | 284 | .alloc_inode = sock_alloc_inode, |
6d7855c5 | 285 | .free_inode = sock_free_inode, |
c6d409cf | 286 | .statfs = simple_statfs, |
1da177e4 LT |
287 | }; |
288 | ||
c23fbb6b ED |
289 | /* |
290 | * sockfs_dname() is called from d_path(). | |
291 | */ | |
292 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
293 | { | |
294 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 295 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
296 | } |
297 | ||
3ba13d17 | 298 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 299 | .d_dname = sockfs_dname, |
1da177e4 LT |
300 | }; |
301 | ||
bba0bd31 AG |
302 | static int sockfs_xattr_get(const struct xattr_handler *handler, |
303 | struct dentry *dentry, struct inode *inode, | |
304 | const char *suffix, void *value, size_t size) | |
305 | { | |
306 | if (value) { | |
307 | if (dentry->d_name.len + 1 > size) | |
308 | return -ERANGE; | |
309 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
310 | } | |
311 | return dentry->d_name.len + 1; | |
312 | } | |
313 | ||
314 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" | |
315 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
316 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
317 | ||
318 | static const struct xattr_handler sockfs_xattr_handler = { | |
319 | .name = XATTR_NAME_SOCKPROTONAME, | |
320 | .get = sockfs_xattr_get, | |
321 | }; | |
322 | ||
4a590153 AG |
323 | static int sockfs_security_xattr_set(const struct xattr_handler *handler, |
324 | struct dentry *dentry, struct inode *inode, | |
325 | const char *suffix, const void *value, | |
326 | size_t size, int flags) | |
327 | { | |
328 | /* Handled by LSM. */ | |
329 | return -EAGAIN; | |
330 | } | |
331 | ||
332 | static const struct xattr_handler sockfs_security_xattr_handler = { | |
333 | .prefix = XATTR_SECURITY_PREFIX, | |
334 | .set = sockfs_security_xattr_set, | |
335 | }; | |
336 | ||
bba0bd31 AG |
337 | static const struct xattr_handler *sockfs_xattr_handlers[] = { |
338 | &sockfs_xattr_handler, | |
4a590153 | 339 | &sockfs_security_xattr_handler, |
bba0bd31 AG |
340 | NULL |
341 | }; | |
342 | ||
fba9be49 | 343 | static int sockfs_init_fs_context(struct fs_context *fc) |
c74a1cbb | 344 | { |
fba9be49 DH |
345 | struct pseudo_fs_context *ctx = init_pseudo(fc, SOCKFS_MAGIC); |
346 | if (!ctx) | |
347 | return -ENOMEM; | |
348 | ctx->ops = &sockfs_ops; | |
349 | ctx->dops = &sockfs_dentry_operations; | |
350 | ctx->xattr = sockfs_xattr_handlers; | |
351 | return 0; | |
c74a1cbb AV |
352 | } |
353 | ||
354 | static struct vfsmount *sock_mnt __read_mostly; | |
355 | ||
356 | static struct file_system_type sock_fs_type = { | |
357 | .name = "sockfs", | |
fba9be49 | 358 | .init_fs_context = sockfs_init_fs_context, |
c74a1cbb AV |
359 | .kill_sb = kill_anon_super, |
360 | }; | |
361 | ||
1da177e4 LT |
362 | /* |
363 | * Obtains the first available file descriptor and sets it up for use. | |
364 | * | |
39d8c1b6 DM |
365 | * These functions create file structures and maps them to fd space |
366 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
367 | * and file struct implicitly stored in sock->file. |
368 | * Note that another thread may close file descriptor before we return | |
369 | * from this function. We use the fact that now we do not refer | |
370 | * to socket after mapping. If one day we will need it, this | |
371 | * function will increment ref. count on file by 1. | |
372 | * | |
373 | * In any case returned fd MAY BE not valid! | |
374 | * This race condition is unavoidable | |
375 | * with shared fd spaces, we cannot solve it inside kernel, | |
376 | * but we take care of internal coherence yet. | |
377 | */ | |
378 | ||
8a3c245c PT |
379 | /** |
380 | * sock_alloc_file - Bind a &socket to a &file | |
381 | * @sock: socket | |
382 | * @flags: file status flags | |
383 | * @dname: protocol name | |
384 | * | |
385 | * Returns the &file bound with @sock, implicitly storing it | |
386 | * in sock->file. If dname is %NULL, sets to "". | |
387 | * On failure the return is a ERR pointer (see linux/err.h). | |
388 | * This function uses GFP_KERNEL internally. | |
389 | */ | |
390 | ||
aab174f0 | 391 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 392 | { |
7cbe66b6 | 393 | struct file *file; |
1da177e4 | 394 | |
d93aa9d8 AV |
395 | if (!dname) |
396 | dname = sock->sk ? sock->sk->sk_prot_creator->name : ""; | |
39d8c1b6 | 397 | |
d93aa9d8 AV |
398 | file = alloc_file_pseudo(SOCK_INODE(sock), sock_mnt, dname, |
399 | O_RDWR | (flags & O_NONBLOCK), | |
400 | &socket_file_ops); | |
b5ffe634 | 401 | if (IS_ERR(file)) { |
8e1611e2 | 402 | sock_release(sock); |
39b65252 | 403 | return file; |
cc3808f8 AV |
404 | } |
405 | ||
406 | sock->file = file; | |
39d8c1b6 | 407 | file->private_data = sock; |
d8e464ec | 408 | stream_open(SOCK_INODE(sock), file); |
28407630 | 409 | return file; |
39d8c1b6 | 410 | } |
56b31d1c | 411 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 412 | |
56b31d1c | 413 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
414 | { |
415 | struct file *newfile; | |
28407630 | 416 | int fd = get_unused_fd_flags(flags); |
ce4bb04c AV |
417 | if (unlikely(fd < 0)) { |
418 | sock_release(sock); | |
28407630 | 419 | return fd; |
ce4bb04c | 420 | } |
39d8c1b6 | 421 | |
aab174f0 | 422 | newfile = sock_alloc_file(sock, flags, NULL); |
4546e44c | 423 | if (!IS_ERR(newfile)) { |
39d8c1b6 | 424 | fd_install(fd, newfile); |
28407630 AV |
425 | return fd; |
426 | } | |
7cbe66b6 | 427 | |
28407630 AV |
428 | put_unused_fd(fd); |
429 | return PTR_ERR(newfile); | |
1da177e4 LT |
430 | } |
431 | ||
8a3c245c PT |
432 | /** |
433 | * sock_from_file - Return the &socket bounded to @file. | |
434 | * @file: file | |
435 | * @err: pointer to an error code return | |
436 | * | |
437 | * On failure returns %NULL and assigns -ENOTSOCK to @err. | |
438 | */ | |
439 | ||
406a3c63 | 440 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 441 | { |
6cb153ca BL |
442 | if (file->f_op == &socket_file_ops) |
443 | return file->private_data; /* set in sock_map_fd */ | |
444 | ||
23bb80d2 ED |
445 | *err = -ENOTSOCK; |
446 | return NULL; | |
6cb153ca | 447 | } |
406a3c63 | 448 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 449 | |
1da177e4 | 450 | /** |
c6d409cf | 451 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
452 | * @fd: file handle |
453 | * @err: pointer to an error code return | |
454 | * | |
455 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 456 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
457 | * with a negative errno code and NULL is returned. The function checks |
458 | * for both invalid handles and passing a handle which is not a socket. | |
459 | * | |
460 | * On a success the socket object pointer is returned. | |
461 | */ | |
462 | ||
463 | struct socket *sockfd_lookup(int fd, int *err) | |
464 | { | |
465 | struct file *file; | |
1da177e4 LT |
466 | struct socket *sock; |
467 | ||
89bddce5 SH |
468 | file = fget(fd); |
469 | if (!file) { | |
1da177e4 LT |
470 | *err = -EBADF; |
471 | return NULL; | |
472 | } | |
89bddce5 | 473 | |
6cb153ca BL |
474 | sock = sock_from_file(file, err); |
475 | if (!sock) | |
1da177e4 | 476 | fput(file); |
6cb153ca BL |
477 | return sock; |
478 | } | |
c6d409cf | 479 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 480 | |
6cb153ca BL |
481 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
482 | { | |
00e188ef | 483 | struct fd f = fdget(fd); |
6cb153ca BL |
484 | struct socket *sock; |
485 | ||
3672558c | 486 | *err = -EBADF; |
00e188ef AV |
487 | if (f.file) { |
488 | sock = sock_from_file(f.file, err); | |
489 | if (likely(sock)) { | |
490 | *fput_needed = f.flags; | |
6cb153ca | 491 | return sock; |
00e188ef AV |
492 | } |
493 | fdput(f); | |
1da177e4 | 494 | } |
6cb153ca | 495 | return NULL; |
1da177e4 LT |
496 | } |
497 | ||
600e1779 MY |
498 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
499 | size_t size) | |
500 | { | |
501 | ssize_t len; | |
502 | ssize_t used = 0; | |
503 | ||
c5ef6035 | 504 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
505 | if (len < 0) |
506 | return len; | |
507 | used += len; | |
508 | if (buffer) { | |
509 | if (size < used) | |
510 | return -ERANGE; | |
511 | buffer += len; | |
512 | } | |
513 | ||
514 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
515 | used += len; | |
516 | if (buffer) { | |
517 | if (size < used) | |
518 | return -ERANGE; | |
519 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
520 | buffer += len; | |
521 | } | |
522 | ||
523 | return used; | |
524 | } | |
525 | ||
dc647ec8 | 526 | static int sockfs_setattr(struct dentry *dentry, struct iattr *iattr) |
86741ec2 LC |
527 | { |
528 | int err = simple_setattr(dentry, iattr); | |
529 | ||
e1a3a60a | 530 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
531 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
532 | ||
6d8c50dc CW |
533 | if (sock->sk) |
534 | sock->sk->sk_uid = iattr->ia_uid; | |
535 | else | |
536 | err = -ENOENT; | |
86741ec2 LC |
537 | } |
538 | ||
539 | return err; | |
540 | } | |
541 | ||
600e1779 | 542 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 543 | .listxattr = sockfs_listxattr, |
86741ec2 | 544 | .setattr = sockfs_setattr, |
600e1779 MY |
545 | }; |
546 | ||
1da177e4 | 547 | /** |
8a3c245c | 548 | * sock_alloc - allocate a socket |
89bddce5 | 549 | * |
1da177e4 LT |
550 | * Allocate a new inode and socket object. The two are bound together |
551 | * and initialised. The socket is then returned. If we are out of inodes | |
8a3c245c | 552 | * NULL is returned. This functions uses GFP_KERNEL internally. |
1da177e4 LT |
553 | */ |
554 | ||
f4a00aac | 555 | struct socket *sock_alloc(void) |
1da177e4 | 556 | { |
89bddce5 SH |
557 | struct inode *inode; |
558 | struct socket *sock; | |
1da177e4 | 559 | |
a209dfc7 | 560 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
561 | if (!inode) |
562 | return NULL; | |
563 | ||
564 | sock = SOCKET_I(inode); | |
565 | ||
85fe4025 | 566 | inode->i_ino = get_next_ino(); |
89bddce5 | 567 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
568 | inode->i_uid = current_fsuid(); |
569 | inode->i_gid = current_fsgid(); | |
600e1779 | 570 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 571 | |
1da177e4 LT |
572 | return sock; |
573 | } | |
f4a00aac | 574 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 575 | |
1da177e4 | 576 | /** |
8a3c245c | 577 | * sock_release - close a socket |
1da177e4 LT |
578 | * @sock: socket to close |
579 | * | |
580 | * The socket is released from the protocol stack if it has a release | |
581 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 582 | * an inode not a file. |
1da177e4 | 583 | */ |
89bddce5 | 584 | |
6d8c50dc | 585 | static void __sock_release(struct socket *sock, struct inode *inode) |
1da177e4 LT |
586 | { |
587 | if (sock->ops) { | |
588 | struct module *owner = sock->ops->owner; | |
589 | ||
6d8c50dc CW |
590 | if (inode) |
591 | inode_lock(inode); | |
1da177e4 | 592 | sock->ops->release(sock); |
ff7b11aa | 593 | sock->sk = NULL; |
6d8c50dc CW |
594 | if (inode) |
595 | inode_unlock(inode); | |
1da177e4 LT |
596 | sock->ops = NULL; |
597 | module_put(owner); | |
598 | } | |
599 | ||
333f7909 | 600 | if (sock->wq.fasync_list) |
3410f22e | 601 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 602 | |
1da177e4 LT |
603 | if (!sock->file) { |
604 | iput(SOCK_INODE(sock)); | |
605 | return; | |
606 | } | |
89bddce5 | 607 | sock->file = NULL; |
1da177e4 | 608 | } |
6d8c50dc CW |
609 | |
610 | void sock_release(struct socket *sock) | |
611 | { | |
612 | __sock_release(sock, NULL); | |
613 | } | |
c6d409cf | 614 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 615 | |
c14ac945 | 616 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 617 | { |
140c55d4 ED |
618 | u8 flags = *tx_flags; |
619 | ||
c14ac945 | 620 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
621 | flags |= SKBTX_HW_TSTAMP; |
622 | ||
c14ac945 | 623 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
624 | flags |= SKBTX_SW_TSTAMP; |
625 | ||
c14ac945 | 626 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
627 | flags |= SKBTX_SCHED_TSTAMP; |
628 | ||
140c55d4 | 629 | *tx_flags = flags; |
20d49473 | 630 | } |
67cc0d40 | 631 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 632 | |
8c3c447b PA |
633 | INDIRECT_CALLABLE_DECLARE(int inet_sendmsg(struct socket *, struct msghdr *, |
634 | size_t)); | |
a648a592 PA |
635 | INDIRECT_CALLABLE_DECLARE(int inet6_sendmsg(struct socket *, struct msghdr *, |
636 | size_t)); | |
d8725c86 | 637 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 638 | { |
a648a592 PA |
639 | int ret = INDIRECT_CALL_INET(sock->ops->sendmsg, inet6_sendmsg, |
640 | inet_sendmsg, sock, msg, | |
641 | msg_data_left(msg)); | |
d8725c86 AV |
642 | BUG_ON(ret == -EIOCBQUEUED); |
643 | return ret; | |
1da177e4 LT |
644 | } |
645 | ||
85806af0 RD |
646 | /** |
647 | * sock_sendmsg - send a message through @sock | |
648 | * @sock: socket | |
649 | * @msg: message to send | |
650 | * | |
651 | * Sends @msg through @sock, passing through LSM. | |
652 | * Returns the number of bytes sent, or an error code. | |
653 | */ | |
d8725c86 | 654 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 655 | { |
d8725c86 | 656 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 657 | msg_data_left(msg)); |
228e548e | 658 | |
d8725c86 | 659 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 660 | } |
c6d409cf | 661 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 | 662 | |
8a3c245c PT |
663 | /** |
664 | * kernel_sendmsg - send a message through @sock (kernel-space) | |
665 | * @sock: socket | |
666 | * @msg: message header | |
667 | * @vec: kernel vec | |
668 | * @num: vec array length | |
669 | * @size: total message data size | |
670 | * | |
671 | * Builds the message data with @vec and sends it through @sock. | |
672 | * Returns the number of bytes sent, or an error code. | |
673 | */ | |
674 | ||
1da177e4 LT |
675 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
676 | struct kvec *vec, size_t num, size_t size) | |
677 | { | |
aa563d7b | 678 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
d8725c86 | 679 | return sock_sendmsg(sock, msg); |
1da177e4 | 680 | } |
c6d409cf | 681 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 682 | |
8a3c245c PT |
683 | /** |
684 | * kernel_sendmsg_locked - send a message through @sock (kernel-space) | |
685 | * @sk: sock | |
686 | * @msg: message header | |
687 | * @vec: output s/g array | |
688 | * @num: output s/g array length | |
689 | * @size: total message data size | |
690 | * | |
691 | * Builds the message data with @vec and sends it through @sock. | |
692 | * Returns the number of bytes sent, or an error code. | |
693 | * Caller must hold @sk. | |
694 | */ | |
695 | ||
306b13eb TH |
696 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
697 | struct kvec *vec, size_t num, size_t size) | |
698 | { | |
699 | struct socket *sock = sk->sk_socket; | |
700 | ||
701 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 702 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb | 703 | |
aa563d7b | 704 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
306b13eb TH |
705 | |
706 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
707 | } | |
708 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
709 | ||
8605330a SHY |
710 | static bool skb_is_err_queue(const struct sk_buff *skb) |
711 | { | |
712 | /* pkt_type of skbs enqueued on the error queue are set to | |
713 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
714 | * in recvmsg, since skbs received on a local socket will never | |
715 | * have a pkt_type of PACKET_OUTGOING. | |
716 | */ | |
717 | return skb->pkt_type == PACKET_OUTGOING; | |
718 | } | |
719 | ||
b50a5c70 ML |
720 | /* On transmit, software and hardware timestamps are returned independently. |
721 | * As the two skb clones share the hardware timestamp, which may be updated | |
722 | * before the software timestamp is received, a hardware TX timestamp may be | |
723 | * returned only if there is no software TX timestamp. Ignore false software | |
724 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
7f1bc6e9 | 725 | * option SO_TIMESTAMP_OLD(NS) is enabled on the socket, even when the skb has a |
b50a5c70 ML |
726 | * hardware timestamp. |
727 | */ | |
728 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
729 | { | |
730 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
731 | } | |
732 | ||
aad9c8c4 ML |
733 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb) |
734 | { | |
735 | struct scm_ts_pktinfo ts_pktinfo; | |
736 | struct net_device *orig_dev; | |
737 | ||
738 | if (!skb_mac_header_was_set(skb)) | |
739 | return; | |
740 | ||
741 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
742 | ||
743 | rcu_read_lock(); | |
744 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
745 | if (orig_dev) | |
746 | ts_pktinfo.if_index = orig_dev->ifindex; | |
747 | rcu_read_unlock(); | |
748 | ||
749 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
750 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
751 | sizeof(ts_pktinfo), &ts_pktinfo); | |
752 | } | |
753 | ||
92f37fd2 ED |
754 | /* |
755 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
756 | */ | |
757 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
758 | struct sk_buff *skb) | |
759 | { | |
20d49473 | 760 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
887feae3 | 761 | int new_tstamp = sock_flag(sk, SOCK_TSTAMP_NEW); |
9718475e DD |
762 | struct scm_timestamping_internal tss; |
763 | ||
b50a5c70 | 764 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
765 | struct skb_shared_hwtstamps *shhwtstamps = |
766 | skb_hwtstamps(skb); | |
767 | ||
768 | /* Race occurred between timestamp enabling and packet | |
769 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 770 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 771 | __net_timestamp(skb); |
b50a5c70 ML |
772 | false_tstamp = 1; |
773 | } | |
20d49473 PO |
774 | |
775 | if (need_software_tstamp) { | |
776 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
887feae3 DD |
777 | if (new_tstamp) { |
778 | struct __kernel_sock_timeval tv; | |
779 | ||
780 | skb_get_new_timestamp(skb, &tv); | |
781 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_NEW, | |
782 | sizeof(tv), &tv); | |
783 | } else { | |
784 | struct __kernel_old_timeval tv; | |
785 | ||
786 | skb_get_timestamp(skb, &tv); | |
787 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_OLD, | |
788 | sizeof(tv), &tv); | |
789 | } | |
20d49473 | 790 | } else { |
887feae3 DD |
791 | if (new_tstamp) { |
792 | struct __kernel_timespec ts; | |
793 | ||
794 | skb_get_new_timestampns(skb, &ts); | |
795 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_NEW, | |
796 | sizeof(ts), &ts); | |
797 | } else { | |
df1b4ba9 | 798 | struct __kernel_old_timespec ts; |
887feae3 DD |
799 | |
800 | skb_get_timestampns(skb, &ts); | |
801 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_OLD, | |
802 | sizeof(ts), &ts); | |
803 | } | |
20d49473 PO |
804 | } |
805 | } | |
806 | ||
f24b9be5 | 807 | memset(&tss, 0, sizeof(tss)); |
c199105d | 808 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
9718475e | 809 | ktime_to_timespec64_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 810 | empty = 0; |
4d276eb6 | 811 | if (shhwtstamps && |
b9f40e21 | 812 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
b50a5c70 | 813 | !skb_is_swtx_tstamp(skb, false_tstamp) && |
9718475e | 814 | ktime_to_timespec64_cond(shhwtstamps->hwtstamp, tss.ts + 2)) { |
4d276eb6 | 815 | empty = 0; |
aad9c8c4 ML |
816 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && |
817 | !skb_is_err_queue(skb)) | |
818 | put_ts_pktinfo(msg, skb); | |
819 | } | |
1c885808 | 820 | if (!empty) { |
9718475e DD |
821 | if (sock_flag(sk, SOCK_TSTAMP_NEW)) |
822 | put_cmsg_scm_timestamping64(msg, &tss); | |
823 | else | |
824 | put_cmsg_scm_timestamping(msg, &tss); | |
1c885808 | 825 | |
8605330a | 826 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 827 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
828 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
829 | skb->len, skb->data); | |
830 | } | |
92f37fd2 | 831 | } |
7c81fd8b ACM |
832 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
833 | ||
6e3e939f JB |
834 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
835 | struct sk_buff *skb) | |
836 | { | |
837 | int ack; | |
838 | ||
839 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
840 | return; | |
841 | if (!skb->wifi_acked_valid) | |
842 | return; | |
843 | ||
844 | ack = skb->wifi_acked; | |
845 | ||
846 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
847 | } | |
848 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
849 | ||
11165f14 | 850 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
851 | struct sk_buff *skb) | |
3b885787 | 852 | { |
744d5a3e | 853 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 854 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 855 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
856 | } |
857 | ||
767dd033 | 858 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
859 | struct sk_buff *skb) |
860 | { | |
861 | sock_recv_timestamp(msg, sk, skb); | |
862 | sock_recv_drops(msg, sk, skb); | |
863 | } | |
767dd033 | 864 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 865 | |
8c3c447b | 866 | INDIRECT_CALLABLE_DECLARE(int inet_recvmsg(struct socket *, struct msghdr *, |
a648a592 PA |
867 | size_t, int)); |
868 | INDIRECT_CALLABLE_DECLARE(int inet6_recvmsg(struct socket *, struct msghdr *, | |
869 | size_t, int)); | |
1b784140 | 870 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 871 | int flags) |
1da177e4 | 872 | { |
a648a592 PA |
873 | return INDIRECT_CALL_INET(sock->ops->recvmsg, inet6_recvmsg, |
874 | inet_recvmsg, sock, msg, msg_data_left(msg), | |
875 | flags); | |
1da177e4 LT |
876 | } |
877 | ||
85806af0 RD |
878 | /** |
879 | * sock_recvmsg - receive a message from @sock | |
880 | * @sock: socket | |
881 | * @msg: message to receive | |
882 | * @flags: message flags | |
883 | * | |
884 | * Receives @msg from @sock, passing through LSM. Returns the total number | |
885 | * of bytes received, or an error. | |
886 | */ | |
2da62906 | 887 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 888 | { |
2da62906 | 889 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 890 | |
2da62906 | 891 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 892 | } |
c6d409cf | 893 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 894 | |
c1249c0a | 895 | /** |
8a3c245c PT |
896 | * kernel_recvmsg - Receive a message from a socket (kernel space) |
897 | * @sock: The socket to receive the message from | |
898 | * @msg: Received message | |
899 | * @vec: Input s/g array for message data | |
900 | * @num: Size of input s/g array | |
901 | * @size: Number of bytes to read | |
902 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
c1249c0a | 903 | * |
8a3c245c PT |
904 | * On return the msg structure contains the scatter/gather array passed in the |
905 | * vec argument. The array is modified so that it consists of the unfilled | |
906 | * portion of the original array. | |
c1249c0a | 907 | * |
8a3c245c | 908 | * The returned value is the total number of bytes received, or an error. |
c1249c0a | 909 | */ |
8a3c245c | 910 | |
89bddce5 SH |
911 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
912 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
913 | { |
914 | mm_segment_t oldfs = get_fs(); | |
915 | int result; | |
916 | ||
aa563d7b | 917 | iov_iter_kvec(&msg->msg_iter, READ, vec, num, size); |
1da177e4 | 918 | set_fs(KERNEL_DS); |
2da62906 | 919 | result = sock_recvmsg(sock, msg, flags); |
1da177e4 LT |
920 | set_fs(oldfs); |
921 | return result; | |
922 | } | |
c6d409cf | 923 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 924 | |
ce1d4d3e CH |
925 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
926 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 927 | { |
1da177e4 LT |
928 | struct socket *sock; |
929 | int flags; | |
930 | ||
ce1d4d3e CH |
931 | sock = file->private_data; |
932 | ||
35f9c09f ED |
933 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
934 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
935 | flags |= more; | |
ce1d4d3e | 936 | |
e6949583 | 937 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 938 | } |
1da177e4 | 939 | |
9c55e01c | 940 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 941 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
942 | unsigned int flags) |
943 | { | |
944 | struct socket *sock = file->private_data; | |
945 | ||
997b37da | 946 | if (unlikely(!sock->ops->splice_read)) |
95506588 | 947 | return generic_file_splice_read(file, ppos, pipe, len, flags); |
997b37da | 948 | |
9c55e01c JA |
949 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
950 | } | |
951 | ||
8ae5e030 | 952 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 953 | { |
6d652330 AV |
954 | struct file *file = iocb->ki_filp; |
955 | struct socket *sock = file->private_data; | |
0345f931 | 956 | struct msghdr msg = {.msg_iter = *to, |
957 | .msg_iocb = iocb}; | |
8ae5e030 | 958 | ssize_t res; |
ce1d4d3e | 959 | |
8ae5e030 AV |
960 | if (file->f_flags & O_NONBLOCK) |
961 | msg.msg_flags = MSG_DONTWAIT; | |
962 | ||
963 | if (iocb->ki_pos != 0) | |
1da177e4 | 964 | return -ESPIPE; |
027445c3 | 965 | |
66ee59af | 966 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
967 | return 0; |
968 | ||
2da62906 | 969 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
970 | *to = msg.msg_iter; |
971 | return res; | |
1da177e4 LT |
972 | } |
973 | ||
8ae5e030 | 974 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 975 | { |
6d652330 AV |
976 | struct file *file = iocb->ki_filp; |
977 | struct socket *sock = file->private_data; | |
0345f931 | 978 | struct msghdr msg = {.msg_iter = *from, |
979 | .msg_iocb = iocb}; | |
8ae5e030 | 980 | ssize_t res; |
1da177e4 | 981 | |
8ae5e030 | 982 | if (iocb->ki_pos != 0) |
ce1d4d3e | 983 | return -ESPIPE; |
027445c3 | 984 | |
8ae5e030 AV |
985 | if (file->f_flags & O_NONBLOCK) |
986 | msg.msg_flags = MSG_DONTWAIT; | |
987 | ||
6d652330 AV |
988 | if (sock->type == SOCK_SEQPACKET) |
989 | msg.msg_flags |= MSG_EOR; | |
990 | ||
d8725c86 | 991 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
992 | *from = msg.msg_iter; |
993 | return res; | |
1da177e4 LT |
994 | } |
995 | ||
1da177e4 LT |
996 | /* |
997 | * Atomic setting of ioctl hooks to avoid race | |
998 | * with module unload. | |
999 | */ | |
1000 | ||
4a3e2f71 | 1001 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 1002 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 1003 | |
881d966b | 1004 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 1005 | { |
4a3e2f71 | 1006 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 1007 | br_ioctl_hook = hook; |
4a3e2f71 | 1008 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1009 | } |
1010 | EXPORT_SYMBOL(brioctl_set); | |
1011 | ||
4a3e2f71 | 1012 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 1013 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 1014 | |
881d966b | 1015 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 1016 | { |
4a3e2f71 | 1017 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1018 | vlan_ioctl_hook = hook; |
4a3e2f71 | 1019 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
1020 | } |
1021 | EXPORT_SYMBOL(vlan_ioctl_set); | |
1022 | ||
4a3e2f71 | 1023 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 1024 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 1025 | |
89bddce5 | 1026 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 1027 | { |
4a3e2f71 | 1028 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 1029 | dlci_ioctl_hook = hook; |
4a3e2f71 | 1030 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
1031 | } |
1032 | EXPORT_SYMBOL(dlci_ioctl_set); | |
1033 | ||
6b96018b | 1034 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
63ff03ab | 1035 | unsigned int cmd, unsigned long arg) |
6b96018b AB |
1036 | { |
1037 | int err; | |
1038 | void __user *argp = (void __user *)arg; | |
1039 | ||
1040 | err = sock->ops->ioctl(sock, cmd, arg); | |
1041 | ||
1042 | /* | |
1043 | * If this ioctl is unknown try to hand it down | |
1044 | * to the NIC driver. | |
1045 | */ | |
36fd633e AV |
1046 | if (err != -ENOIOCTLCMD) |
1047 | return err; | |
6b96018b | 1048 | |
36fd633e AV |
1049 | if (cmd == SIOCGIFCONF) { |
1050 | struct ifconf ifc; | |
1051 | if (copy_from_user(&ifc, argp, sizeof(struct ifconf))) | |
1052 | return -EFAULT; | |
1053 | rtnl_lock(); | |
1054 | err = dev_ifconf(net, &ifc, sizeof(struct ifreq)); | |
1055 | rtnl_unlock(); | |
1056 | if (!err && copy_to_user(argp, &ifc, sizeof(struct ifconf))) | |
1057 | err = -EFAULT; | |
44c02a2c AV |
1058 | } else { |
1059 | struct ifreq ifr; | |
1060 | bool need_copyout; | |
63ff03ab | 1061 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) |
44c02a2c AV |
1062 | return -EFAULT; |
1063 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1064 | if (!err && need_copyout) | |
63ff03ab | 1065 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) |
44c02a2c | 1066 | return -EFAULT; |
36fd633e | 1067 | } |
6b96018b AB |
1068 | return err; |
1069 | } | |
1070 | ||
1da177e4 LT |
1071 | /* |
1072 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1073 | * what to do with it - that's up to the protocol still. | |
1074 | */ | |
1075 | ||
8a3c245c PT |
1076 | /** |
1077 | * get_net_ns - increment the refcount of the network namespace | |
1078 | * @ns: common namespace (net) | |
1079 | * | |
1080 | * Returns the net's common namespace. | |
1081 | */ | |
1082 | ||
d8d211a2 | 1083 | struct ns_common *get_net_ns(struct ns_common *ns) |
c62cce2c AV |
1084 | { |
1085 | return &get_net(container_of(ns, struct net, ns))->ns; | |
1086 | } | |
d8d211a2 | 1087 | EXPORT_SYMBOL_GPL(get_net_ns); |
c62cce2c | 1088 | |
1da177e4 LT |
1089 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
1090 | { | |
1091 | struct socket *sock; | |
881d966b | 1092 | struct sock *sk; |
1da177e4 LT |
1093 | void __user *argp = (void __user *)arg; |
1094 | int pid, err; | |
881d966b | 1095 | struct net *net; |
1da177e4 | 1096 | |
b69aee04 | 1097 | sock = file->private_data; |
881d966b | 1098 | sk = sock->sk; |
3b1e0a65 | 1099 | net = sock_net(sk); |
44c02a2c AV |
1100 | if (unlikely(cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))) { |
1101 | struct ifreq ifr; | |
1102 | bool need_copyout; | |
1103 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
1104 | return -EFAULT; | |
1105 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1106 | if (!err && need_copyout) | |
1107 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
1108 | return -EFAULT; | |
1da177e4 | 1109 | } else |
3d23e349 | 1110 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1111 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
b1b0c245 | 1112 | err = wext_handle_ioctl(net, cmd, argp); |
1da177e4 | 1113 | } else |
3d23e349 | 1114 | #endif |
89bddce5 | 1115 | switch (cmd) { |
1da177e4 LT |
1116 | case FIOSETOWN: |
1117 | case SIOCSPGRP: | |
1118 | err = -EFAULT; | |
1119 | if (get_user(pid, (int __user *)argp)) | |
1120 | break; | |
393cc3f5 | 1121 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1122 | break; |
1123 | case FIOGETOWN: | |
1124 | case SIOCGPGRP: | |
609d7fa9 | 1125 | err = put_user(f_getown(sock->file), |
89bddce5 | 1126 | (int __user *)argp); |
1da177e4 LT |
1127 | break; |
1128 | case SIOCGIFBR: | |
1129 | case SIOCSIFBR: | |
1130 | case SIOCBRADDBR: | |
1131 | case SIOCBRDELBR: | |
1132 | err = -ENOPKG; | |
1133 | if (!br_ioctl_hook) | |
1134 | request_module("bridge"); | |
1135 | ||
4a3e2f71 | 1136 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1137 | if (br_ioctl_hook) |
881d966b | 1138 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1139 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1140 | break; |
1141 | case SIOCGIFVLAN: | |
1142 | case SIOCSIFVLAN: | |
1143 | err = -ENOPKG; | |
1144 | if (!vlan_ioctl_hook) | |
1145 | request_module("8021q"); | |
1146 | ||
4a3e2f71 | 1147 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1148 | if (vlan_ioctl_hook) |
881d966b | 1149 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1150 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1151 | break; |
1da177e4 LT |
1152 | case SIOCADDDLCI: |
1153 | case SIOCDELDLCI: | |
1154 | err = -ENOPKG; | |
1155 | if (!dlci_ioctl_hook) | |
1156 | request_module("dlci"); | |
1157 | ||
7512cbf6 PE |
1158 | mutex_lock(&dlci_ioctl_mutex); |
1159 | if (dlci_ioctl_hook) | |
1da177e4 | 1160 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1161 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 | 1162 | break; |
c62cce2c AV |
1163 | case SIOCGSKNS: |
1164 | err = -EPERM; | |
1165 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1166 | break; | |
1167 | ||
1168 | err = open_related_ns(&net->ns, get_net_ns); | |
1169 | break; | |
0768e170 AB |
1170 | case SIOCGSTAMP_OLD: |
1171 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
1172 | if (!sock->ops->gettstamp) { |
1173 | err = -ENOIOCTLCMD; | |
1174 | break; | |
1175 | } | |
1176 | err = sock->ops->gettstamp(sock, argp, | |
0768e170 AB |
1177 | cmd == SIOCGSTAMP_OLD, |
1178 | !IS_ENABLED(CONFIG_64BIT)); | |
60747828 | 1179 | break; |
0768e170 AB |
1180 | case SIOCGSTAMP_NEW: |
1181 | case SIOCGSTAMPNS_NEW: | |
1182 | if (!sock->ops->gettstamp) { | |
1183 | err = -ENOIOCTLCMD; | |
1184 | break; | |
1185 | } | |
1186 | err = sock->ops->gettstamp(sock, argp, | |
1187 | cmd == SIOCGSTAMP_NEW, | |
1188 | false); | |
c7cbdbf2 | 1189 | break; |
1da177e4 | 1190 | default: |
63ff03ab | 1191 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1192 | break; |
89bddce5 | 1193 | } |
1da177e4 LT |
1194 | return err; |
1195 | } | |
1196 | ||
8a3c245c PT |
1197 | /** |
1198 | * sock_create_lite - creates a socket | |
1199 | * @family: protocol family (AF_INET, ...) | |
1200 | * @type: communication type (SOCK_STREAM, ...) | |
1201 | * @protocol: protocol (0, ...) | |
1202 | * @res: new socket | |
1203 | * | |
1204 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1205 | * The new socket initialization is not complete, see kernel_accept(). | |
1206 | * Returns 0 or an error. On failure @res is set to %NULL. | |
1207 | * This function internally uses GFP_KERNEL. | |
1208 | */ | |
1209 | ||
1da177e4 LT |
1210 | int sock_create_lite(int family, int type, int protocol, struct socket **res) |
1211 | { | |
1212 | int err; | |
1213 | struct socket *sock = NULL; | |
89bddce5 | 1214 | |
1da177e4 LT |
1215 | err = security_socket_create(family, type, protocol, 1); |
1216 | if (err) | |
1217 | goto out; | |
1218 | ||
1219 | sock = sock_alloc(); | |
1220 | if (!sock) { | |
1221 | err = -ENOMEM; | |
1222 | goto out; | |
1223 | } | |
1224 | ||
1da177e4 | 1225 | sock->type = type; |
7420ed23 VY |
1226 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1227 | if (err) | |
1228 | goto out_release; | |
1229 | ||
1da177e4 LT |
1230 | out: |
1231 | *res = sock; | |
1232 | return err; | |
7420ed23 VY |
1233 | out_release: |
1234 | sock_release(sock); | |
1235 | sock = NULL; | |
1236 | goto out; | |
1da177e4 | 1237 | } |
c6d409cf | 1238 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1239 | |
1240 | /* No kernel lock held - perfect */ | |
ade994f4 | 1241 | static __poll_t sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1242 | { |
3cafb376 | 1243 | struct socket *sock = file->private_data; |
a331de3b | 1244 | __poll_t events = poll_requested_events(wait), flag = 0; |
2d48d67f | 1245 | |
e88958e6 CH |
1246 | if (!sock->ops->poll) |
1247 | return 0; | |
f641f13b | 1248 | |
a331de3b CH |
1249 | if (sk_can_busy_loop(sock->sk)) { |
1250 | /* poll once if requested by the syscall */ | |
1251 | if (events & POLL_BUSY_LOOP) | |
1252 | sk_busy_loop(sock->sk, 1); | |
1253 | ||
1254 | /* if this socket can poll_ll, tell the system call */ | |
1255 | flag = POLL_BUSY_LOOP; | |
1256 | } | |
1257 | ||
1258 | return sock->ops->poll(file, sock, wait) | flag; | |
1da177e4 LT |
1259 | } |
1260 | ||
89bddce5 | 1261 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1262 | { |
b69aee04 | 1263 | struct socket *sock = file->private_data; |
1da177e4 LT |
1264 | |
1265 | return sock->ops->mmap(file, sock, vma); | |
1266 | } | |
1267 | ||
20380731 | 1268 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1269 | { |
6d8c50dc | 1270 | __sock_release(SOCKET_I(inode), inode); |
1da177e4 LT |
1271 | return 0; |
1272 | } | |
1273 | ||
1274 | /* | |
1275 | * Update the socket async list | |
1276 | * | |
1277 | * Fasync_list locking strategy. | |
1278 | * | |
1279 | * 1. fasync_list is modified only under process context socket lock | |
1280 | * i.e. under semaphore. | |
1281 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1282 | * or under socket lock |
1da177e4 LT |
1283 | */ |
1284 | ||
1285 | static int sock_fasync(int fd, struct file *filp, int on) | |
1286 | { | |
989a2979 ED |
1287 | struct socket *sock = filp->private_data; |
1288 | struct sock *sk = sock->sk; | |
333f7909 | 1289 | struct socket_wq *wq = &sock->wq; |
1da177e4 | 1290 | |
989a2979 | 1291 | if (sk == NULL) |
1da177e4 | 1292 | return -EINVAL; |
1da177e4 LT |
1293 | |
1294 | lock_sock(sk); | |
eaefd110 | 1295 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1296 | |
eaefd110 | 1297 | if (!wq->fasync_list) |
989a2979 ED |
1298 | sock_reset_flag(sk, SOCK_FASYNC); |
1299 | else | |
bcdce719 | 1300 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1301 | |
989a2979 | 1302 | release_sock(sk); |
1da177e4 LT |
1303 | return 0; |
1304 | } | |
1305 | ||
ceb5d58b | 1306 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1307 | |
ceb5d58b | 1308 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1309 | { |
ceb5d58b | 1310 | if (!wq || !wq->fasync_list) |
1da177e4 | 1311 | return -1; |
ceb5d58b | 1312 | |
89bddce5 | 1313 | switch (how) { |
8d8ad9d7 | 1314 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1315 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1316 | break; |
1317 | goto call_kill; | |
8d8ad9d7 | 1318 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1319 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 LT |
1320 | break; |
1321 | /* fall through */ | |
8d8ad9d7 | 1322 | case SOCK_WAKE_IO: |
89bddce5 | 1323 | call_kill: |
43815482 | 1324 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1325 | break; |
8d8ad9d7 | 1326 | case SOCK_WAKE_URG: |
43815482 | 1327 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1328 | } |
ceb5d58b | 1329 | |
1da177e4 LT |
1330 | return 0; |
1331 | } | |
c6d409cf | 1332 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1333 | |
8a3c245c PT |
1334 | /** |
1335 | * __sock_create - creates a socket | |
1336 | * @net: net namespace | |
1337 | * @family: protocol family (AF_INET, ...) | |
1338 | * @type: communication type (SOCK_STREAM, ...) | |
1339 | * @protocol: protocol (0, ...) | |
1340 | * @res: new socket | |
1341 | * @kern: boolean for kernel space sockets | |
1342 | * | |
1343 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1344 | * Returns 0 or an error. On failure @res is set to %NULL. @kern must | |
1345 | * be set to true if the socket resides in kernel space. | |
1346 | * This function internally uses GFP_KERNEL. | |
1347 | */ | |
1348 | ||
721db93a | 1349 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1350 | struct socket **res, int kern) |
1da177e4 LT |
1351 | { |
1352 | int err; | |
1353 | struct socket *sock; | |
55737fda | 1354 | const struct net_proto_family *pf; |
1da177e4 LT |
1355 | |
1356 | /* | |
89bddce5 | 1357 | * Check protocol is in range |
1da177e4 LT |
1358 | */ |
1359 | if (family < 0 || family >= NPROTO) | |
1360 | return -EAFNOSUPPORT; | |
1361 | if (type < 0 || type >= SOCK_MAX) | |
1362 | return -EINVAL; | |
1363 | ||
1364 | /* Compatibility. | |
1365 | ||
1366 | This uglymoron is moved from INET layer to here to avoid | |
1367 | deadlock in module load. | |
1368 | */ | |
1369 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1370 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1371 | current->comm); | |
1da177e4 LT |
1372 | family = PF_PACKET; |
1373 | } | |
1374 | ||
1375 | err = security_socket_create(family, type, protocol, kern); | |
1376 | if (err) | |
1377 | return err; | |
89bddce5 | 1378 | |
55737fda SH |
1379 | /* |
1380 | * Allocate the socket and allow the family to set things up. if | |
1381 | * the protocol is 0, the family is instructed to select an appropriate | |
1382 | * default. | |
1383 | */ | |
1384 | sock = sock_alloc(); | |
1385 | if (!sock) { | |
e87cc472 | 1386 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1387 | return -ENFILE; /* Not exactly a match, but its the |
1388 | closest posix thing */ | |
1389 | } | |
1390 | ||
1391 | sock->type = type; | |
1392 | ||
95a5afca | 1393 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1394 | /* Attempt to load a protocol module if the find failed. |
1395 | * | |
1396 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1397 | * requested real, full-featured networking support upon configuration. |
1398 | * Otherwise module support will break! | |
1399 | */ | |
190683a9 | 1400 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1401 | request_module("net-pf-%d", family); |
1da177e4 LT |
1402 | #endif |
1403 | ||
55737fda SH |
1404 | rcu_read_lock(); |
1405 | pf = rcu_dereference(net_families[family]); | |
1406 | err = -EAFNOSUPPORT; | |
1407 | if (!pf) | |
1408 | goto out_release; | |
1da177e4 LT |
1409 | |
1410 | /* | |
1411 | * We will call the ->create function, that possibly is in a loadable | |
1412 | * module, so we have to bump that loadable module refcnt first. | |
1413 | */ | |
55737fda | 1414 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1415 | goto out_release; |
1416 | ||
55737fda SH |
1417 | /* Now protected by module ref count */ |
1418 | rcu_read_unlock(); | |
1419 | ||
3f378b68 | 1420 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1421 | if (err < 0) |
1da177e4 | 1422 | goto out_module_put; |
a79af59e | 1423 | |
1da177e4 LT |
1424 | /* |
1425 | * Now to bump the refcnt of the [loadable] module that owns this | |
1426 | * socket at sock_release time we decrement its refcnt. | |
1427 | */ | |
55737fda SH |
1428 | if (!try_module_get(sock->ops->owner)) |
1429 | goto out_module_busy; | |
1430 | ||
1da177e4 LT |
1431 | /* |
1432 | * Now that we're done with the ->create function, the [loadable] | |
1433 | * module can have its refcnt decremented | |
1434 | */ | |
55737fda | 1435 | module_put(pf->owner); |
7420ed23 VY |
1436 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1437 | if (err) | |
3b185525 | 1438 | goto out_sock_release; |
55737fda | 1439 | *res = sock; |
1da177e4 | 1440 | |
55737fda SH |
1441 | return 0; |
1442 | ||
1443 | out_module_busy: | |
1444 | err = -EAFNOSUPPORT; | |
1da177e4 | 1445 | out_module_put: |
55737fda SH |
1446 | sock->ops = NULL; |
1447 | module_put(pf->owner); | |
1448 | out_sock_release: | |
1da177e4 | 1449 | sock_release(sock); |
55737fda SH |
1450 | return err; |
1451 | ||
1452 | out_release: | |
1453 | rcu_read_unlock(); | |
1454 | goto out_sock_release; | |
1da177e4 | 1455 | } |
721db93a | 1456 | EXPORT_SYMBOL(__sock_create); |
1da177e4 | 1457 | |
8a3c245c PT |
1458 | /** |
1459 | * sock_create - creates a socket | |
1460 | * @family: protocol family (AF_INET, ...) | |
1461 | * @type: communication type (SOCK_STREAM, ...) | |
1462 | * @protocol: protocol (0, ...) | |
1463 | * @res: new socket | |
1464 | * | |
1465 | * A wrapper around __sock_create(). | |
1466 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1467 | */ | |
1468 | ||
1da177e4 LT |
1469 | int sock_create(int family, int type, int protocol, struct socket **res) |
1470 | { | |
1b8d7ae4 | 1471 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1472 | } |
c6d409cf | 1473 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1474 | |
8a3c245c PT |
1475 | /** |
1476 | * sock_create_kern - creates a socket (kernel space) | |
1477 | * @net: net namespace | |
1478 | * @family: protocol family (AF_INET, ...) | |
1479 | * @type: communication type (SOCK_STREAM, ...) | |
1480 | * @protocol: protocol (0, ...) | |
1481 | * @res: new socket | |
1482 | * | |
1483 | * A wrapper around __sock_create(). | |
1484 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1485 | */ | |
1486 | ||
eeb1bd5c | 1487 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1488 | { |
eeb1bd5c | 1489 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1490 | } |
c6d409cf | 1491 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1492 | |
9d6a15c3 | 1493 | int __sys_socket(int family, int type, int protocol) |
1da177e4 LT |
1494 | { |
1495 | int retval; | |
1496 | struct socket *sock; | |
a677a039 UD |
1497 | int flags; |
1498 | ||
e38b36f3 UD |
1499 | /* Check the SOCK_* constants for consistency. */ |
1500 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1501 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1502 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1503 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1504 | ||
a677a039 | 1505 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1506 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1507 | return -EINVAL; |
1508 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1509 | |
aaca0bdc UD |
1510 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1511 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1512 | ||
1da177e4 LT |
1513 | retval = sock_create(family, type, protocol, &sock); |
1514 | if (retval < 0) | |
8e1611e2 | 1515 | return retval; |
1da177e4 | 1516 | |
8e1611e2 | 1517 | return sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1518 | } |
1519 | ||
9d6a15c3 DB |
1520 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1521 | { | |
1522 | return __sys_socket(family, type, protocol); | |
1523 | } | |
1524 | ||
1da177e4 LT |
1525 | /* |
1526 | * Create a pair of connected sockets. | |
1527 | */ | |
1528 | ||
6debc8d8 | 1529 | int __sys_socketpair(int family, int type, int protocol, int __user *usockvec) |
1da177e4 LT |
1530 | { |
1531 | struct socket *sock1, *sock2; | |
1532 | int fd1, fd2, err; | |
db349509 | 1533 | struct file *newfile1, *newfile2; |
a677a039 UD |
1534 | int flags; |
1535 | ||
1536 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1537 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1538 | return -EINVAL; |
1539 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1540 | |
aaca0bdc UD |
1541 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1542 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1543 | ||
016a266b AV |
1544 | /* |
1545 | * reserve descriptors and make sure we won't fail | |
1546 | * to return them to userland. | |
1547 | */ | |
1548 | fd1 = get_unused_fd_flags(flags); | |
1549 | if (unlikely(fd1 < 0)) | |
1550 | return fd1; | |
1551 | ||
1552 | fd2 = get_unused_fd_flags(flags); | |
1553 | if (unlikely(fd2 < 0)) { | |
1554 | put_unused_fd(fd1); | |
1555 | return fd2; | |
1556 | } | |
1557 | ||
1558 | err = put_user(fd1, &usockvec[0]); | |
1559 | if (err) | |
1560 | goto out; | |
1561 | ||
1562 | err = put_user(fd2, &usockvec[1]); | |
1563 | if (err) | |
1564 | goto out; | |
1565 | ||
1da177e4 LT |
1566 | /* |
1567 | * Obtain the first socket and check if the underlying protocol | |
1568 | * supports the socketpair call. | |
1569 | */ | |
1570 | ||
1571 | err = sock_create(family, type, protocol, &sock1); | |
016a266b | 1572 | if (unlikely(err < 0)) |
1da177e4 LT |
1573 | goto out; |
1574 | ||
1575 | err = sock_create(family, type, protocol, &sock2); | |
016a266b AV |
1576 | if (unlikely(err < 0)) { |
1577 | sock_release(sock1); | |
1578 | goto out; | |
bf3c23d1 | 1579 | } |
d73aa286 | 1580 | |
d47cd945 DH |
1581 | err = security_socket_socketpair(sock1, sock2); |
1582 | if (unlikely(err)) { | |
1583 | sock_release(sock2); | |
1584 | sock_release(sock1); | |
1585 | goto out; | |
1586 | } | |
1587 | ||
016a266b AV |
1588 | err = sock1->ops->socketpair(sock1, sock2); |
1589 | if (unlikely(err < 0)) { | |
1590 | sock_release(sock2); | |
1591 | sock_release(sock1); | |
1592 | goto out; | |
28407630 AV |
1593 | } |
1594 | ||
aab174f0 | 1595 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1596 | if (IS_ERR(newfile1)) { |
28407630 | 1597 | err = PTR_ERR(newfile1); |
016a266b AV |
1598 | sock_release(sock2); |
1599 | goto out; | |
28407630 AV |
1600 | } |
1601 | ||
aab174f0 | 1602 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1603 | if (IS_ERR(newfile2)) { |
1604 | err = PTR_ERR(newfile2); | |
016a266b AV |
1605 | fput(newfile1); |
1606 | goto out; | |
db349509 AV |
1607 | } |
1608 | ||
157cf649 | 1609 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1610 | |
db349509 AV |
1611 | fd_install(fd1, newfile1); |
1612 | fd_install(fd2, newfile2); | |
d73aa286 | 1613 | return 0; |
1da177e4 | 1614 | |
016a266b | 1615 | out: |
d73aa286 | 1616 | put_unused_fd(fd2); |
d73aa286 | 1617 | put_unused_fd(fd1); |
1da177e4 LT |
1618 | return err; |
1619 | } | |
1620 | ||
6debc8d8 DB |
1621 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1622 | int __user *, usockvec) | |
1623 | { | |
1624 | return __sys_socketpair(family, type, protocol, usockvec); | |
1625 | } | |
1626 | ||
1da177e4 LT |
1627 | /* |
1628 | * Bind a name to a socket. Nothing much to do here since it's | |
1629 | * the protocol's responsibility to handle the local address. | |
1630 | * | |
1631 | * We move the socket address to kernel space before we call | |
1632 | * the protocol layer (having also checked the address is ok). | |
1633 | */ | |
1634 | ||
a87d35d8 | 1635 | int __sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
1da177e4 LT |
1636 | { |
1637 | struct socket *sock; | |
230b1839 | 1638 | struct sockaddr_storage address; |
6cb153ca | 1639 | int err, fput_needed; |
1da177e4 | 1640 | |
89bddce5 | 1641 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1642 | if (sock) { |
43db362d | 1643 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
068b88cc | 1644 | if (!err) { |
89bddce5 | 1645 | err = security_socket_bind(sock, |
230b1839 | 1646 | (struct sockaddr *)&address, |
89bddce5 | 1647 | addrlen); |
6cb153ca BL |
1648 | if (!err) |
1649 | err = sock->ops->bind(sock, | |
89bddce5 | 1650 | (struct sockaddr *) |
230b1839 | 1651 | &address, addrlen); |
1da177e4 | 1652 | } |
6cb153ca | 1653 | fput_light(sock->file, fput_needed); |
89bddce5 | 1654 | } |
1da177e4 LT |
1655 | return err; |
1656 | } | |
1657 | ||
a87d35d8 DB |
1658 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1659 | { | |
1660 | return __sys_bind(fd, umyaddr, addrlen); | |
1661 | } | |
1662 | ||
1da177e4 LT |
1663 | /* |
1664 | * Perform a listen. Basically, we allow the protocol to do anything | |
1665 | * necessary for a listen, and if that works, we mark the socket as | |
1666 | * ready for listening. | |
1667 | */ | |
1668 | ||
25e290ee | 1669 | int __sys_listen(int fd, int backlog) |
1da177e4 LT |
1670 | { |
1671 | struct socket *sock; | |
6cb153ca | 1672 | int err, fput_needed; |
b8e1f9b5 | 1673 | int somaxconn; |
89bddce5 SH |
1674 | |
1675 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1676 | if (sock) { | |
8efa6e93 | 1677 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1678 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1679 | backlog = somaxconn; |
1da177e4 LT |
1680 | |
1681 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1682 | if (!err) |
1683 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1684 | |
6cb153ca | 1685 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1686 | } |
1687 | return err; | |
1688 | } | |
1689 | ||
25e290ee DB |
1690 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1691 | { | |
1692 | return __sys_listen(fd, backlog); | |
1693 | } | |
1694 | ||
de2ea4b6 JA |
1695 | int __sys_accept4_file(struct file *file, unsigned file_flags, |
1696 | struct sockaddr __user *upeer_sockaddr, | |
1697 | int __user *upeer_addrlen, int flags) | |
1da177e4 LT |
1698 | { |
1699 | struct socket *sock, *newsock; | |
39d8c1b6 | 1700 | struct file *newfile; |
de2ea4b6 | 1701 | int err, len, newfd; |
230b1839 | 1702 | struct sockaddr_storage address; |
1da177e4 | 1703 | |
77d27200 | 1704 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1705 | return -EINVAL; |
1706 | ||
1707 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1708 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1709 | ||
de2ea4b6 | 1710 | sock = sock_from_file(file, &err); |
1da177e4 LT |
1711 | if (!sock) |
1712 | goto out; | |
1713 | ||
1714 | err = -ENFILE; | |
c6d409cf ED |
1715 | newsock = sock_alloc(); |
1716 | if (!newsock) | |
de2ea4b6 | 1717 | goto out; |
1da177e4 LT |
1718 | |
1719 | newsock->type = sock->type; | |
1720 | newsock->ops = sock->ops; | |
1721 | ||
1da177e4 LT |
1722 | /* |
1723 | * We don't need try_module_get here, as the listening socket (sock) | |
1724 | * has the protocol module (sock->ops->owner) held. | |
1725 | */ | |
1726 | __module_get(newsock->ops->owner); | |
1727 | ||
28407630 | 1728 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1729 | if (unlikely(newfd < 0)) { |
1730 | err = newfd; | |
9a1875e6 | 1731 | sock_release(newsock); |
de2ea4b6 | 1732 | goto out; |
39d8c1b6 | 1733 | } |
aab174f0 | 1734 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1735 | if (IS_ERR(newfile)) { |
28407630 AV |
1736 | err = PTR_ERR(newfile); |
1737 | put_unused_fd(newfd); | |
de2ea4b6 | 1738 | goto out; |
28407630 | 1739 | } |
39d8c1b6 | 1740 | |
a79af59e FF |
1741 | err = security_socket_accept(sock, newsock); |
1742 | if (err) | |
39d8c1b6 | 1743 | goto out_fd; |
a79af59e | 1744 | |
de2ea4b6 JA |
1745 | err = sock->ops->accept(sock, newsock, sock->file->f_flags | file_flags, |
1746 | false); | |
1da177e4 | 1747 | if (err < 0) |
39d8c1b6 | 1748 | goto out_fd; |
1da177e4 LT |
1749 | |
1750 | if (upeer_sockaddr) { | |
9b2c45d4 DV |
1751 | len = newsock->ops->getname(newsock, |
1752 | (struct sockaddr *)&address, 2); | |
1753 | if (len < 0) { | |
1da177e4 | 1754 | err = -ECONNABORTED; |
39d8c1b6 | 1755 | goto out_fd; |
1da177e4 | 1756 | } |
43db362d | 1757 | err = move_addr_to_user(&address, |
230b1839 | 1758 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1759 | if (err < 0) |
39d8c1b6 | 1760 | goto out_fd; |
1da177e4 LT |
1761 | } |
1762 | ||
1763 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1764 | ||
39d8c1b6 DM |
1765 | fd_install(newfd, newfile); |
1766 | err = newfd; | |
1da177e4 LT |
1767 | out: |
1768 | return err; | |
39d8c1b6 | 1769 | out_fd: |
9606a216 | 1770 | fput(newfile); |
39d8c1b6 | 1771 | put_unused_fd(newfd); |
de2ea4b6 JA |
1772 | goto out; |
1773 | ||
1774 | } | |
1775 | ||
1776 | /* | |
1777 | * For accept, we attempt to create a new socket, set up the link | |
1778 | * with the client, wake up the client, then return the new | |
1779 | * connected fd. We collect the address of the connector in kernel | |
1780 | * space and move it to user at the very end. This is unclean because | |
1781 | * we open the socket then return an error. | |
1782 | * | |
1783 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1784 | * status to recvmsg. We need to add that support in a way thats | |
1785 | * clean when we restructure accept also. | |
1786 | */ | |
1787 | ||
1788 | int __sys_accept4(int fd, struct sockaddr __user *upeer_sockaddr, | |
1789 | int __user *upeer_addrlen, int flags) | |
1790 | { | |
1791 | int ret = -EBADF; | |
1792 | struct fd f; | |
1793 | ||
1794 | f = fdget(fd); | |
1795 | if (f.file) { | |
1796 | ret = __sys_accept4_file(f.file, 0, upeer_sockaddr, | |
1797 | upeer_addrlen, flags); | |
1798 | if (f.flags) | |
1799 | fput(f.file); | |
1800 | } | |
1801 | ||
1802 | return ret; | |
1da177e4 LT |
1803 | } |
1804 | ||
4541e805 DB |
1805 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1806 | int __user *, upeer_addrlen, int, flags) | |
1807 | { | |
1808 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, flags); | |
1809 | } | |
1810 | ||
20f37034 HC |
1811 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1812 | int __user *, upeer_addrlen) | |
aaca0bdc | 1813 | { |
4541e805 | 1814 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1815 | } |
1816 | ||
1da177e4 LT |
1817 | /* |
1818 | * Attempt to connect to a socket with the server address. The address | |
1819 | * is in user space so we verify it is OK and move it to kernel space. | |
1820 | * | |
1821 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1822 | * break bindings | |
1823 | * | |
1824 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1825 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1826 | * include the -EINPROGRESS status for such sockets. | |
1827 | */ | |
1828 | ||
bd3ded31 JA |
1829 | int __sys_connect_file(struct file *file, struct sockaddr __user *uservaddr, |
1830 | int addrlen, int file_flags) | |
1da177e4 LT |
1831 | { |
1832 | struct socket *sock; | |
230b1839 | 1833 | struct sockaddr_storage address; |
bd3ded31 | 1834 | int err; |
1da177e4 | 1835 | |
bd3ded31 | 1836 | sock = sock_from_file(file, &err); |
1da177e4 LT |
1837 | if (!sock) |
1838 | goto out; | |
43db362d | 1839 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 | 1840 | if (err < 0) |
bd3ded31 | 1841 | goto out; |
1da177e4 | 1842 | |
89bddce5 | 1843 | err = |
230b1839 | 1844 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 | 1845 | if (err) |
bd3ded31 | 1846 | goto out; |
1da177e4 | 1847 | |
230b1839 | 1848 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
bd3ded31 | 1849 | sock->file->f_flags | file_flags); |
1da177e4 LT |
1850 | out: |
1851 | return err; | |
1852 | } | |
1853 | ||
bd3ded31 JA |
1854 | int __sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
1855 | { | |
1856 | int ret = -EBADF; | |
1857 | struct fd f; | |
1858 | ||
1859 | f = fdget(fd); | |
1860 | if (f.file) { | |
1861 | ret = __sys_connect_file(f.file, uservaddr, addrlen, 0); | |
1862 | if (f.flags) | |
1863 | fput(f.file); | |
1864 | } | |
1865 | ||
1866 | return ret; | |
1867 | } | |
1868 | ||
1387c2c2 DB |
1869 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1870 | int, addrlen) | |
1871 | { | |
1872 | return __sys_connect(fd, uservaddr, addrlen); | |
1873 | } | |
1874 | ||
1da177e4 LT |
1875 | /* |
1876 | * Get the local address ('name') of a socket object. Move the obtained | |
1877 | * name to user space. | |
1878 | */ | |
1879 | ||
8882a107 DB |
1880 | int __sys_getsockname(int fd, struct sockaddr __user *usockaddr, |
1881 | int __user *usockaddr_len) | |
1da177e4 LT |
1882 | { |
1883 | struct socket *sock; | |
230b1839 | 1884 | struct sockaddr_storage address; |
9b2c45d4 | 1885 | int err, fput_needed; |
89bddce5 | 1886 | |
6cb153ca | 1887 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1888 | if (!sock) |
1889 | goto out; | |
1890 | ||
1891 | err = security_socket_getsockname(sock); | |
1892 | if (err) | |
1893 | goto out_put; | |
1894 | ||
9b2c45d4 DV |
1895 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 0); |
1896 | if (err < 0) | |
1da177e4 | 1897 | goto out_put; |
9b2c45d4 DV |
1898 | /* "err" is actually length in this case */ |
1899 | err = move_addr_to_user(&address, err, usockaddr, usockaddr_len); | |
1da177e4 LT |
1900 | |
1901 | out_put: | |
6cb153ca | 1902 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1903 | out: |
1904 | return err; | |
1905 | } | |
1906 | ||
8882a107 DB |
1907 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1908 | int __user *, usockaddr_len) | |
1909 | { | |
1910 | return __sys_getsockname(fd, usockaddr, usockaddr_len); | |
1911 | } | |
1912 | ||
1da177e4 LT |
1913 | /* |
1914 | * Get the remote address ('name') of a socket object. Move the obtained | |
1915 | * name to user space. | |
1916 | */ | |
1917 | ||
b21c8f83 DB |
1918 | int __sys_getpeername(int fd, struct sockaddr __user *usockaddr, |
1919 | int __user *usockaddr_len) | |
1da177e4 LT |
1920 | { |
1921 | struct socket *sock; | |
230b1839 | 1922 | struct sockaddr_storage address; |
9b2c45d4 | 1923 | int err, fput_needed; |
1da177e4 | 1924 | |
89bddce5 SH |
1925 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1926 | if (sock != NULL) { | |
1da177e4 LT |
1927 | err = security_socket_getpeername(sock); |
1928 | if (err) { | |
6cb153ca | 1929 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1930 | return err; |
1931 | } | |
1932 | ||
9b2c45d4 DV |
1933 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 1); |
1934 | if (err >= 0) | |
1935 | /* "err" is actually length in this case */ | |
1936 | err = move_addr_to_user(&address, err, usockaddr, | |
89bddce5 | 1937 | usockaddr_len); |
6cb153ca | 1938 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1939 | } |
1940 | return err; | |
1941 | } | |
1942 | ||
b21c8f83 DB |
1943 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1944 | int __user *, usockaddr_len) | |
1945 | { | |
1946 | return __sys_getpeername(fd, usockaddr, usockaddr_len); | |
1947 | } | |
1948 | ||
1da177e4 LT |
1949 | /* |
1950 | * Send a datagram to a given address. We move the address into kernel | |
1951 | * space and check the user space data area is readable before invoking | |
1952 | * the protocol. | |
1953 | */ | |
211b634b DB |
1954 | int __sys_sendto(int fd, void __user *buff, size_t len, unsigned int flags, |
1955 | struct sockaddr __user *addr, int addr_len) | |
1da177e4 LT |
1956 | { |
1957 | struct socket *sock; | |
230b1839 | 1958 | struct sockaddr_storage address; |
1da177e4 LT |
1959 | int err; |
1960 | struct msghdr msg; | |
1961 | struct iovec iov; | |
6cb153ca | 1962 | int fput_needed; |
6cb153ca | 1963 | |
602bd0e9 AV |
1964 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1965 | if (unlikely(err)) | |
1966 | return err; | |
de0fa95c PE |
1967 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1968 | if (!sock) | |
4387ff75 | 1969 | goto out; |
6cb153ca | 1970 | |
89bddce5 | 1971 | msg.msg_name = NULL; |
89bddce5 SH |
1972 | msg.msg_control = NULL; |
1973 | msg.msg_controllen = 0; | |
1974 | msg.msg_namelen = 0; | |
6cb153ca | 1975 | if (addr) { |
43db362d | 1976 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1977 | if (err < 0) |
1978 | goto out_put; | |
230b1839 | 1979 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1980 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1981 | } |
1982 | if (sock->file->f_flags & O_NONBLOCK) | |
1983 | flags |= MSG_DONTWAIT; | |
1984 | msg.msg_flags = flags; | |
d8725c86 | 1985 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 1986 | |
89bddce5 | 1987 | out_put: |
de0fa95c | 1988 | fput_light(sock->file, fput_needed); |
4387ff75 | 1989 | out: |
1da177e4 LT |
1990 | return err; |
1991 | } | |
1992 | ||
211b634b DB |
1993 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
1994 | unsigned int, flags, struct sockaddr __user *, addr, | |
1995 | int, addr_len) | |
1996 | { | |
1997 | return __sys_sendto(fd, buff, len, flags, addr, addr_len); | |
1998 | } | |
1999 | ||
1da177e4 | 2000 | /* |
89bddce5 | 2001 | * Send a datagram down a socket. |
1da177e4 LT |
2002 | */ |
2003 | ||
3e0fa65f | 2004 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 2005 | unsigned int, flags) |
1da177e4 | 2006 | { |
211b634b | 2007 | return __sys_sendto(fd, buff, len, flags, NULL, 0); |
1da177e4 LT |
2008 | } |
2009 | ||
2010 | /* | |
89bddce5 | 2011 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
2012 | * sender. We verify the buffers are writable and if needed move the |
2013 | * sender address from kernel to user space. | |
2014 | */ | |
7a09e1eb DB |
2015 | int __sys_recvfrom(int fd, void __user *ubuf, size_t size, unsigned int flags, |
2016 | struct sockaddr __user *addr, int __user *addr_len) | |
1da177e4 LT |
2017 | { |
2018 | struct socket *sock; | |
2019 | struct iovec iov; | |
2020 | struct msghdr msg; | |
230b1839 | 2021 | struct sockaddr_storage address; |
89bddce5 | 2022 | int err, err2; |
6cb153ca BL |
2023 | int fput_needed; |
2024 | ||
602bd0e9 AV |
2025 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
2026 | if (unlikely(err)) | |
2027 | return err; | |
de0fa95c | 2028 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 2029 | if (!sock) |
de0fa95c | 2030 | goto out; |
1da177e4 | 2031 | |
89bddce5 SH |
2032 | msg.msg_control = NULL; |
2033 | msg.msg_controllen = 0; | |
f3d33426 HFS |
2034 | /* Save some cycles and don't copy the address if not needed */ |
2035 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
2036 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
2037 | msg.msg_namelen = 0; | |
130ed5d1 | 2038 | msg.msg_iocb = NULL; |
9f138fa6 | 2039 | msg.msg_flags = 0; |
1da177e4 LT |
2040 | if (sock->file->f_flags & O_NONBLOCK) |
2041 | flags |= MSG_DONTWAIT; | |
2da62906 | 2042 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 2043 | |
89bddce5 | 2044 | if (err >= 0 && addr != NULL) { |
43db362d | 2045 | err2 = move_addr_to_user(&address, |
230b1839 | 2046 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
2047 | if (err2 < 0) |
2048 | err = err2; | |
1da177e4 | 2049 | } |
de0fa95c PE |
2050 | |
2051 | fput_light(sock->file, fput_needed); | |
4387ff75 | 2052 | out: |
1da177e4 LT |
2053 | return err; |
2054 | } | |
2055 | ||
7a09e1eb DB |
2056 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
2057 | unsigned int, flags, struct sockaddr __user *, addr, | |
2058 | int __user *, addr_len) | |
2059 | { | |
2060 | return __sys_recvfrom(fd, ubuf, size, flags, addr, addr_len); | |
2061 | } | |
2062 | ||
1da177e4 | 2063 | /* |
89bddce5 | 2064 | * Receive a datagram from a socket. |
1da177e4 LT |
2065 | */ |
2066 | ||
b7c0ddf5 JG |
2067 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
2068 | unsigned int, flags) | |
1da177e4 | 2069 | { |
7a09e1eb | 2070 | return __sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
1da177e4 LT |
2071 | } |
2072 | ||
2073 | /* | |
2074 | * Set a socket option. Because we don't know the option lengths we have | |
2075 | * to pass the user mode parameter for the protocols to sort out. | |
2076 | */ | |
2077 | ||
cc36dca0 DB |
2078 | static int __sys_setsockopt(int fd, int level, int optname, |
2079 | char __user *optval, int optlen) | |
1da177e4 | 2080 | { |
0d01da6a SF |
2081 | mm_segment_t oldfs = get_fs(); |
2082 | char *kernel_optval = NULL; | |
6cb153ca | 2083 | int err, fput_needed; |
1da177e4 LT |
2084 | struct socket *sock; |
2085 | ||
2086 | if (optlen < 0) | |
2087 | return -EINVAL; | |
89bddce5 SH |
2088 | |
2089 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2090 | if (sock != NULL) { | |
2091 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
2092 | if (err) |
2093 | goto out_put; | |
1da177e4 | 2094 | |
0d01da6a SF |
2095 | err = BPF_CGROUP_RUN_PROG_SETSOCKOPT(sock->sk, &level, |
2096 | &optname, optval, &optlen, | |
2097 | &kernel_optval); | |
2098 | ||
2099 | if (err < 0) { | |
2100 | goto out_put; | |
2101 | } else if (err > 0) { | |
2102 | err = 0; | |
2103 | goto out_put; | |
2104 | } | |
2105 | ||
2106 | if (kernel_optval) { | |
2107 | set_fs(KERNEL_DS); | |
2108 | optval = (char __user __force *)kernel_optval; | |
2109 | } | |
2110 | ||
1da177e4 | 2111 | if (level == SOL_SOCKET) |
89bddce5 SH |
2112 | err = |
2113 | sock_setsockopt(sock, level, optname, optval, | |
2114 | optlen); | |
1da177e4 | 2115 | else |
89bddce5 SH |
2116 | err = |
2117 | sock->ops->setsockopt(sock, level, optname, optval, | |
2118 | optlen); | |
0d01da6a SF |
2119 | |
2120 | if (kernel_optval) { | |
2121 | set_fs(oldfs); | |
2122 | kfree(kernel_optval); | |
2123 | } | |
6cb153ca BL |
2124 | out_put: |
2125 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2126 | } |
2127 | return err; | |
2128 | } | |
2129 | ||
cc36dca0 DB |
2130 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
2131 | char __user *, optval, int, optlen) | |
2132 | { | |
2133 | return __sys_setsockopt(fd, level, optname, optval, optlen); | |
2134 | } | |
2135 | ||
1da177e4 LT |
2136 | /* |
2137 | * Get a socket option. Because we don't know the option lengths we have | |
2138 | * to pass a user mode parameter for the protocols to sort out. | |
2139 | */ | |
2140 | ||
13a2d70e DB |
2141 | static int __sys_getsockopt(int fd, int level, int optname, |
2142 | char __user *optval, int __user *optlen) | |
1da177e4 | 2143 | { |
6cb153ca | 2144 | int err, fput_needed; |
1da177e4 | 2145 | struct socket *sock; |
0d01da6a | 2146 | int max_optlen; |
1da177e4 | 2147 | |
89bddce5 SH |
2148 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2149 | if (sock != NULL) { | |
6cb153ca BL |
2150 | err = security_socket_getsockopt(sock, level, optname); |
2151 | if (err) | |
2152 | goto out_put; | |
1da177e4 | 2153 | |
0d01da6a SF |
2154 | max_optlen = BPF_CGROUP_GETSOCKOPT_MAX_OPTLEN(optlen); |
2155 | ||
1da177e4 | 2156 | if (level == SOL_SOCKET) |
89bddce5 SH |
2157 | err = |
2158 | sock_getsockopt(sock, level, optname, optval, | |
2159 | optlen); | |
1da177e4 | 2160 | else |
89bddce5 SH |
2161 | err = |
2162 | sock->ops->getsockopt(sock, level, optname, optval, | |
2163 | optlen); | |
0d01da6a SF |
2164 | |
2165 | err = BPF_CGROUP_RUN_PROG_GETSOCKOPT(sock->sk, level, optname, | |
2166 | optval, optlen, | |
2167 | max_optlen, err); | |
6cb153ca BL |
2168 | out_put: |
2169 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2170 | } |
2171 | return err; | |
2172 | } | |
2173 | ||
13a2d70e DB |
2174 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
2175 | char __user *, optval, int __user *, optlen) | |
2176 | { | |
2177 | return __sys_getsockopt(fd, level, optname, optval, optlen); | |
2178 | } | |
2179 | ||
1da177e4 LT |
2180 | /* |
2181 | * Shutdown a socket. | |
2182 | */ | |
2183 | ||
005a1aea | 2184 | int __sys_shutdown(int fd, int how) |
1da177e4 | 2185 | { |
6cb153ca | 2186 | int err, fput_needed; |
1da177e4 LT |
2187 | struct socket *sock; |
2188 | ||
89bddce5 SH |
2189 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2190 | if (sock != NULL) { | |
1da177e4 | 2191 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
2192 | if (!err) |
2193 | err = sock->ops->shutdown(sock, how); | |
2194 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2195 | } |
2196 | return err; | |
2197 | } | |
2198 | ||
005a1aea DB |
2199 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
2200 | { | |
2201 | return __sys_shutdown(fd, how); | |
2202 | } | |
2203 | ||
89bddce5 | 2204 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
2205 | * fields which are the same type (int / unsigned) on our platforms. |
2206 | */ | |
2207 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
2208 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
2209 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
2210 | ||
c71d8ebe TH |
2211 | struct used_address { |
2212 | struct sockaddr_storage name; | |
2213 | unsigned int name_len; | |
2214 | }; | |
2215 | ||
da184284 AV |
2216 | static int copy_msghdr_from_user(struct msghdr *kmsg, |
2217 | struct user_msghdr __user *umsg, | |
2218 | struct sockaddr __user **save_addr, | |
2219 | struct iovec **iov) | |
1661bf36 | 2220 | { |
ffb07550 | 2221 | struct user_msghdr msg; |
08adb7da AV |
2222 | ssize_t err; |
2223 | ||
ffb07550 | 2224 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 2225 | return -EFAULT; |
dbb490b9 | 2226 | |
864d9664 | 2227 | kmsg->msg_control = (void __force *)msg.msg_control; |
ffb07550 AV |
2228 | kmsg->msg_controllen = msg.msg_controllen; |
2229 | kmsg->msg_flags = msg.msg_flags; | |
2230 | ||
2231 | kmsg->msg_namelen = msg.msg_namelen; | |
2232 | if (!msg.msg_name) | |
6a2a2b3a AS |
2233 | kmsg->msg_namelen = 0; |
2234 | ||
dbb490b9 ML |
2235 | if (kmsg->msg_namelen < 0) |
2236 | return -EINVAL; | |
2237 | ||
1661bf36 | 2238 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 2239 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
2240 | |
2241 | if (save_addr) | |
ffb07550 | 2242 | *save_addr = msg.msg_name; |
08adb7da | 2243 | |
ffb07550 | 2244 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 2245 | if (!save_addr) { |
864d9664 PA |
2246 | err = move_addr_to_kernel(msg.msg_name, |
2247 | kmsg->msg_namelen, | |
08adb7da AV |
2248 | kmsg->msg_name); |
2249 | if (err < 0) | |
2250 | return err; | |
2251 | } | |
2252 | } else { | |
2253 | kmsg->msg_name = NULL; | |
2254 | kmsg->msg_namelen = 0; | |
2255 | } | |
2256 | ||
ffb07550 | 2257 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
2258 | return -EMSGSIZE; |
2259 | ||
0345f931 | 2260 | kmsg->msg_iocb = NULL; |
2261 | ||
87e5e6da | 2262 | err = import_iovec(save_addr ? READ : WRITE, |
ffb07550 | 2263 | msg.msg_iov, msg.msg_iovlen, |
da184284 | 2264 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
87e5e6da | 2265 | return err < 0 ? err : 0; |
1661bf36 DC |
2266 | } |
2267 | ||
4257c8ca JA |
2268 | static int ____sys_sendmsg(struct socket *sock, struct msghdr *msg_sys, |
2269 | unsigned int flags, struct used_address *used_address, | |
2270 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 2271 | { |
b9d717a7 | 2272 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 2273 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 2274 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 2275 | unsigned char *ctl_buf = ctl; |
d8725c86 | 2276 | int ctl_len; |
08adb7da | 2277 | ssize_t err; |
89bddce5 | 2278 | |
1da177e4 LT |
2279 | err = -ENOBUFS; |
2280 | ||
228e548e | 2281 | if (msg_sys->msg_controllen > INT_MAX) |
4257c8ca | 2282 | goto out; |
28a94d8f | 2283 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2284 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2285 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2286 | err = |
228e548e | 2287 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2288 | sizeof(ctl)); |
1da177e4 | 2289 | if (err) |
4257c8ca | 2290 | goto out; |
228e548e AB |
2291 | ctl_buf = msg_sys->msg_control; |
2292 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2293 | } else if (ctl_len) { |
ac4340fc DM |
2294 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2295 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2296 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2297 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2298 | if (ctl_buf == NULL) |
4257c8ca | 2299 | goto out; |
1da177e4 LT |
2300 | } |
2301 | err = -EFAULT; | |
2302 | /* | |
228e548e | 2303 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2304 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2305 | * checking falls down on this. | |
2306 | */ | |
fb8621bb | 2307 | if (copy_from_user(ctl_buf, |
228e548e | 2308 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2309 | ctl_len)) |
1da177e4 | 2310 | goto out_freectl; |
228e548e | 2311 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2312 | } |
228e548e | 2313 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2314 | |
2315 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2316 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2317 | /* |
2318 | * If this is sendmmsg() and current destination address is same as | |
2319 | * previously succeeded address, omit asking LSM's decision. | |
2320 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2321 | * destination address never matches. | |
2322 | */ | |
bc909d9d MD |
2323 | if (used_address && msg_sys->msg_name && |
2324 | used_address->name_len == msg_sys->msg_namelen && | |
2325 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2326 | used_address->name_len)) { |
d8725c86 | 2327 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2328 | goto out_freectl; |
2329 | } | |
d8725c86 | 2330 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2331 | /* |
2332 | * If this is sendmmsg() and sending to current destination address was | |
2333 | * successful, remember it. | |
2334 | */ | |
2335 | if (used_address && err >= 0) { | |
2336 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2337 | if (msg_sys->msg_name) |
2338 | memcpy(&used_address->name, msg_sys->msg_name, | |
2339 | used_address->name_len); | |
c71d8ebe | 2340 | } |
1da177e4 LT |
2341 | |
2342 | out_freectl: | |
89bddce5 | 2343 | if (ctl_buf != ctl) |
1da177e4 | 2344 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
4257c8ca JA |
2345 | out: |
2346 | return err; | |
2347 | } | |
2348 | ||
2349 | static int sendmsg_copy_msghdr(struct msghdr *msg, | |
2350 | struct user_msghdr __user *umsg, unsigned flags, | |
2351 | struct iovec **iov) | |
2352 | { | |
2353 | int err; | |
2354 | ||
2355 | if (flags & MSG_CMSG_COMPAT) { | |
2356 | struct compat_msghdr __user *msg_compat; | |
2357 | ||
2358 | msg_compat = (struct compat_msghdr __user *) umsg; | |
2359 | err = get_compat_msghdr(msg, msg_compat, NULL, iov); | |
2360 | } else { | |
2361 | err = copy_msghdr_from_user(msg, umsg, NULL, iov); | |
2362 | } | |
2363 | if (err < 0) | |
2364 | return err; | |
2365 | ||
2366 | return 0; | |
2367 | } | |
2368 | ||
2369 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, | |
2370 | struct msghdr *msg_sys, unsigned int flags, | |
2371 | struct used_address *used_address, | |
2372 | unsigned int allowed_msghdr_flags) | |
2373 | { | |
2374 | struct sockaddr_storage address; | |
2375 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; | |
2376 | ssize_t err; | |
2377 | ||
2378 | msg_sys->msg_name = &address; | |
2379 | ||
2380 | err = sendmsg_copy_msghdr(msg_sys, msg, flags, &iov); | |
2381 | if (err < 0) | |
2382 | return err; | |
2383 | ||
2384 | err = ____sys_sendmsg(sock, msg_sys, flags, used_address, | |
2385 | allowed_msghdr_flags); | |
da184284 | 2386 | kfree(iov); |
228e548e AB |
2387 | return err; |
2388 | } | |
2389 | ||
2390 | /* | |
2391 | * BSD sendmsg interface | |
2392 | */ | |
d69e0779 | 2393 | long __sys_sendmsg_sock(struct socket *sock, struct user_msghdr __user *umsg, |
0fa03c62 JA |
2394 | unsigned int flags) |
2395 | { | |
d69e0779 JA |
2396 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
2397 | struct sockaddr_storage address; | |
2398 | struct msghdr msg = { .msg_name = &address }; | |
2399 | ssize_t err; | |
0fa03c62 | 2400 | |
d69e0779 JA |
2401 | err = sendmsg_copy_msghdr(&msg, umsg, flags, &iov); |
2402 | if (err) | |
2403 | return err; | |
2404 | /* disallow ancillary data requests from this path */ | |
2405 | if (msg.msg_control || msg.msg_controllen) { | |
2406 | err = -EINVAL; | |
2407 | goto out; | |
2408 | } | |
2409 | ||
2410 | err = ____sys_sendmsg(sock, &msg, flags, NULL, 0); | |
2411 | out: | |
2412 | kfree(iov); | |
2413 | return err; | |
0fa03c62 | 2414 | } |
228e548e | 2415 | |
e1834a32 DB |
2416 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2417 | bool forbid_cmsg_compat) | |
228e548e AB |
2418 | { |
2419 | int fput_needed, err; | |
2420 | struct msghdr msg_sys; | |
1be374a0 AL |
2421 | struct socket *sock; |
2422 | ||
e1834a32 DB |
2423 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2424 | return -EINVAL; | |
2425 | ||
1be374a0 | 2426 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2427 | if (!sock) |
2428 | goto out; | |
2429 | ||
28a94d8f | 2430 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2431 | |
6cb153ca | 2432 | fput_light(sock->file, fput_needed); |
89bddce5 | 2433 | out: |
1da177e4 LT |
2434 | return err; |
2435 | } | |
2436 | ||
666547ff | 2437 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 | 2438 | { |
e1834a32 | 2439 | return __sys_sendmsg(fd, msg, flags, true); |
a7526eb5 AL |
2440 | } |
2441 | ||
228e548e AB |
2442 | /* |
2443 | * Linux sendmmsg interface | |
2444 | */ | |
2445 | ||
2446 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
e1834a32 | 2447 | unsigned int flags, bool forbid_cmsg_compat) |
228e548e AB |
2448 | { |
2449 | int fput_needed, err, datagrams; | |
2450 | struct socket *sock; | |
2451 | struct mmsghdr __user *entry; | |
2452 | struct compat_mmsghdr __user *compat_entry; | |
2453 | struct msghdr msg_sys; | |
c71d8ebe | 2454 | struct used_address used_address; |
f092276d | 2455 | unsigned int oflags = flags; |
228e548e | 2456 | |
e1834a32 DB |
2457 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2458 | return -EINVAL; | |
2459 | ||
98382f41 AB |
2460 | if (vlen > UIO_MAXIOV) |
2461 | vlen = UIO_MAXIOV; | |
228e548e AB |
2462 | |
2463 | datagrams = 0; | |
2464 | ||
2465 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2466 | if (!sock) | |
2467 | return err; | |
2468 | ||
c71d8ebe | 2469 | used_address.name_len = UINT_MAX; |
228e548e AB |
2470 | entry = mmsg; |
2471 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2472 | err = 0; |
f092276d | 2473 | flags |= MSG_BATCH; |
228e548e AB |
2474 | |
2475 | while (datagrams < vlen) { | |
f092276d TH |
2476 | if (datagrams == vlen - 1) |
2477 | flags = oflags; | |
2478 | ||
228e548e | 2479 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2480 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2481 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2482 | if (err < 0) |
2483 | break; | |
2484 | err = __put_user(err, &compat_entry->msg_len); | |
2485 | ++compat_entry; | |
2486 | } else { | |
a7526eb5 | 2487 | err = ___sys_sendmsg(sock, |
666547ff | 2488 | (struct user_msghdr __user *)entry, |
28a94d8f | 2489 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2490 | if (err < 0) |
2491 | break; | |
2492 | err = put_user(err, &entry->msg_len); | |
2493 | ++entry; | |
2494 | } | |
2495 | ||
2496 | if (err) | |
2497 | break; | |
2498 | ++datagrams; | |
3023898b SHY |
2499 | if (msg_data_left(&msg_sys)) |
2500 | break; | |
a78cb84c | 2501 | cond_resched(); |
228e548e AB |
2502 | } |
2503 | ||
228e548e AB |
2504 | fput_light(sock->file, fput_needed); |
2505 | ||
728ffb86 AB |
2506 | /* We only return an error if no datagrams were able to be sent */ |
2507 | if (datagrams != 0) | |
228e548e AB |
2508 | return datagrams; |
2509 | ||
228e548e AB |
2510 | return err; |
2511 | } | |
2512 | ||
2513 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2514 | unsigned int, vlen, unsigned int, flags) | |
2515 | { | |
e1834a32 | 2516 | return __sys_sendmmsg(fd, mmsg, vlen, flags, true); |
228e548e AB |
2517 | } |
2518 | ||
4257c8ca JA |
2519 | static int recvmsg_copy_msghdr(struct msghdr *msg, |
2520 | struct user_msghdr __user *umsg, unsigned flags, | |
2521 | struct sockaddr __user **uaddr, | |
2522 | struct iovec **iov) | |
1da177e4 | 2523 | { |
08adb7da | 2524 | ssize_t err; |
1da177e4 | 2525 | |
4257c8ca JA |
2526 | if (MSG_CMSG_COMPAT & flags) { |
2527 | struct compat_msghdr __user *msg_compat; | |
1da177e4 | 2528 | |
4257c8ca JA |
2529 | msg_compat = (struct compat_msghdr __user *) umsg; |
2530 | err = get_compat_msghdr(msg, msg_compat, uaddr, iov); | |
2531 | } else { | |
2532 | err = copy_msghdr_from_user(msg, umsg, uaddr, iov); | |
2533 | } | |
1da177e4 | 2534 | if (err < 0) |
da184284 | 2535 | return err; |
1da177e4 | 2536 | |
4257c8ca JA |
2537 | return 0; |
2538 | } | |
2539 | ||
2540 | static int ____sys_recvmsg(struct socket *sock, struct msghdr *msg_sys, | |
2541 | struct user_msghdr __user *msg, | |
2542 | struct sockaddr __user *uaddr, | |
2543 | unsigned int flags, int nosec) | |
2544 | { | |
2545 | struct compat_msghdr __user *msg_compat = | |
2546 | (struct compat_msghdr __user *) msg; | |
2547 | int __user *uaddr_len = COMPAT_NAMELEN(msg); | |
2548 | struct sockaddr_storage addr; | |
2549 | unsigned long cmsg_ptr; | |
2550 | int len; | |
2551 | ssize_t err; | |
2552 | ||
2553 | msg_sys->msg_name = &addr; | |
a2e27255 ACM |
2554 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2555 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2556 | |
f3d33426 HFS |
2557 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2558 | msg_sys->msg_namelen = 0; | |
2559 | ||
1da177e4 LT |
2560 | if (sock->file->f_flags & O_NONBLOCK) |
2561 | flags |= MSG_DONTWAIT; | |
2da62906 | 2562 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, flags); |
1da177e4 | 2563 | if (err < 0) |
4257c8ca | 2564 | goto out; |
1da177e4 LT |
2565 | len = err; |
2566 | ||
2567 | if (uaddr != NULL) { | |
43db362d | 2568 | err = move_addr_to_user(&addr, |
a2e27255 | 2569 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2570 | uaddr_len); |
1da177e4 | 2571 | if (err < 0) |
4257c8ca | 2572 | goto out; |
1da177e4 | 2573 | } |
a2e27255 | 2574 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2575 | COMPAT_FLAGS(msg)); |
1da177e4 | 2576 | if (err) |
4257c8ca | 2577 | goto out; |
1da177e4 | 2578 | if (MSG_CMSG_COMPAT & flags) |
a2e27255 | 2579 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2580 | &msg_compat->msg_controllen); |
2581 | else | |
a2e27255 | 2582 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2583 | &msg->msg_controllen); |
2584 | if (err) | |
4257c8ca | 2585 | goto out; |
1da177e4 | 2586 | err = len; |
4257c8ca JA |
2587 | out: |
2588 | return err; | |
2589 | } | |
2590 | ||
2591 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, | |
2592 | struct msghdr *msg_sys, unsigned int flags, int nosec) | |
2593 | { | |
2594 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; | |
2595 | /* user mode address pointers */ | |
2596 | struct sockaddr __user *uaddr; | |
2597 | ssize_t err; | |
2598 | ||
2599 | err = recvmsg_copy_msghdr(msg_sys, msg, flags, &uaddr, &iov); | |
2600 | if (err < 0) | |
2601 | return err; | |
1da177e4 | 2602 | |
4257c8ca | 2603 | err = ____sys_recvmsg(sock, msg_sys, msg, uaddr, flags, nosec); |
da184284 | 2604 | kfree(iov); |
a2e27255 ACM |
2605 | return err; |
2606 | } | |
2607 | ||
2608 | /* | |
2609 | * BSD recvmsg interface | |
2610 | */ | |
2611 | ||
d69e0779 | 2612 | long __sys_recvmsg_sock(struct socket *sock, struct user_msghdr __user *umsg, |
aa1fa28f JA |
2613 | unsigned int flags) |
2614 | { | |
d69e0779 JA |
2615 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
2616 | struct sockaddr_storage address; | |
2617 | struct msghdr msg = { .msg_name = &address }; | |
2618 | struct sockaddr __user *uaddr; | |
2619 | ssize_t err; | |
2620 | ||
2621 | err = recvmsg_copy_msghdr(&msg, umsg, flags, &uaddr, &iov); | |
2622 | if (err) | |
2623 | return err; | |
2624 | /* disallow ancillary data requests from this path */ | |
2625 | if (msg.msg_control || msg.msg_controllen) { | |
2626 | err = -EINVAL; | |
2627 | goto out; | |
2628 | } | |
aa1fa28f | 2629 | |
d69e0779 JA |
2630 | err = ____sys_recvmsg(sock, &msg, umsg, uaddr, flags, 0); |
2631 | out: | |
2632 | kfree(iov); | |
2633 | return err; | |
aa1fa28f JA |
2634 | } |
2635 | ||
e1834a32 DB |
2636 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2637 | bool forbid_cmsg_compat) | |
a2e27255 ACM |
2638 | { |
2639 | int fput_needed, err; | |
2640 | struct msghdr msg_sys; | |
1be374a0 AL |
2641 | struct socket *sock; |
2642 | ||
e1834a32 DB |
2643 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2644 | return -EINVAL; | |
2645 | ||
1be374a0 | 2646 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2647 | if (!sock) |
2648 | goto out; | |
2649 | ||
a7526eb5 | 2650 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2651 | |
6cb153ca | 2652 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2653 | out: |
2654 | return err; | |
2655 | } | |
2656 | ||
666547ff | 2657 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2658 | unsigned int, flags) |
2659 | { | |
e1834a32 | 2660 | return __sys_recvmsg(fd, msg, flags, true); |
a7526eb5 AL |
2661 | } |
2662 | ||
a2e27255 ACM |
2663 | /* |
2664 | * Linux recvmmsg interface | |
2665 | */ | |
2666 | ||
e11d4284 AB |
2667 | static int do_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2668 | unsigned int vlen, unsigned int flags, | |
2669 | struct timespec64 *timeout) | |
a2e27255 ACM |
2670 | { |
2671 | int fput_needed, err, datagrams; | |
2672 | struct socket *sock; | |
2673 | struct mmsghdr __user *entry; | |
d7256d0e | 2674 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2675 | struct msghdr msg_sys; |
766b9f92 DD |
2676 | struct timespec64 end_time; |
2677 | struct timespec64 timeout64; | |
a2e27255 ACM |
2678 | |
2679 | if (timeout && | |
2680 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2681 | timeout->tv_nsec)) | |
2682 | return -EINVAL; | |
2683 | ||
2684 | datagrams = 0; | |
2685 | ||
2686 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2687 | if (!sock) | |
2688 | return err; | |
2689 | ||
7797dc41 SHY |
2690 | if (likely(!(flags & MSG_ERRQUEUE))) { |
2691 | err = sock_error(sock->sk); | |
2692 | if (err) { | |
2693 | datagrams = err; | |
2694 | goto out_put; | |
2695 | } | |
e623a9e9 | 2696 | } |
a2e27255 ACM |
2697 | |
2698 | entry = mmsg; | |
d7256d0e | 2699 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2700 | |
2701 | while (datagrams < vlen) { | |
2702 | /* | |
2703 | * No need to ask LSM for more than the first datagram. | |
2704 | */ | |
d7256d0e | 2705 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2706 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2707 | &msg_sys, flags & ~MSG_WAITFORONE, |
2708 | datagrams); | |
d7256d0e JMG |
2709 | if (err < 0) |
2710 | break; | |
2711 | err = __put_user(err, &compat_entry->msg_len); | |
2712 | ++compat_entry; | |
2713 | } else { | |
a7526eb5 | 2714 | err = ___sys_recvmsg(sock, |
666547ff | 2715 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2716 | &msg_sys, flags & ~MSG_WAITFORONE, |
2717 | datagrams); | |
d7256d0e JMG |
2718 | if (err < 0) |
2719 | break; | |
2720 | err = put_user(err, &entry->msg_len); | |
2721 | ++entry; | |
2722 | } | |
2723 | ||
a2e27255 ACM |
2724 | if (err) |
2725 | break; | |
a2e27255 ACM |
2726 | ++datagrams; |
2727 | ||
71c5c159 BB |
2728 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2729 | if (flags & MSG_WAITFORONE) | |
2730 | flags |= MSG_DONTWAIT; | |
2731 | ||
a2e27255 | 2732 | if (timeout) { |
766b9f92 | 2733 | ktime_get_ts64(&timeout64); |
c2e6c856 | 2734 | *timeout = timespec64_sub(end_time, timeout64); |
a2e27255 ACM |
2735 | if (timeout->tv_sec < 0) { |
2736 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2737 | break; | |
2738 | } | |
2739 | ||
2740 | /* Timeout, return less than vlen datagrams */ | |
2741 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2742 | break; | |
2743 | } | |
2744 | ||
2745 | /* Out of band data, return right away */ | |
2746 | if (msg_sys.msg_flags & MSG_OOB) | |
2747 | break; | |
a78cb84c | 2748 | cond_resched(); |
a2e27255 ACM |
2749 | } |
2750 | ||
a2e27255 | 2751 | if (err == 0) |
34b88a68 ACM |
2752 | goto out_put; |
2753 | ||
2754 | if (datagrams == 0) { | |
2755 | datagrams = err; | |
2756 | goto out_put; | |
2757 | } | |
a2e27255 | 2758 | |
34b88a68 ACM |
2759 | /* |
2760 | * We may return less entries than requested (vlen) if the | |
2761 | * sock is non block and there aren't enough datagrams... | |
2762 | */ | |
2763 | if (err != -EAGAIN) { | |
a2e27255 | 2764 | /* |
34b88a68 ACM |
2765 | * ... or if recvmsg returns an error after we |
2766 | * received some datagrams, where we record the | |
2767 | * error to return on the next call or if the | |
2768 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2769 | */ |
34b88a68 | 2770 | sock->sk->sk_err = -err; |
a2e27255 | 2771 | } |
34b88a68 ACM |
2772 | out_put: |
2773 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2774 | |
34b88a68 | 2775 | return datagrams; |
a2e27255 ACM |
2776 | } |
2777 | ||
e11d4284 AB |
2778 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2779 | unsigned int vlen, unsigned int flags, | |
2780 | struct __kernel_timespec __user *timeout, | |
2781 | struct old_timespec32 __user *timeout32) | |
a2e27255 ACM |
2782 | { |
2783 | int datagrams; | |
c2e6c856 | 2784 | struct timespec64 timeout_sys; |
a2e27255 | 2785 | |
e11d4284 AB |
2786 | if (timeout && get_timespec64(&timeout_sys, timeout)) |
2787 | return -EFAULT; | |
a2e27255 | 2788 | |
e11d4284 | 2789 | if (timeout32 && get_old_timespec32(&timeout_sys, timeout32)) |
a2e27255 ACM |
2790 | return -EFAULT; |
2791 | ||
e11d4284 AB |
2792 | if (!timeout && !timeout32) |
2793 | return do_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2794 | ||
2795 | datagrams = do_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
a2e27255 | 2796 | |
e11d4284 AB |
2797 | if (datagrams <= 0) |
2798 | return datagrams; | |
2799 | ||
2800 | if (timeout && put_timespec64(&timeout_sys, timeout)) | |
2801 | datagrams = -EFAULT; | |
2802 | ||
2803 | if (timeout32 && put_old_timespec32(&timeout_sys, timeout32)) | |
a2e27255 ACM |
2804 | datagrams = -EFAULT; |
2805 | ||
2806 | return datagrams; | |
2807 | } | |
2808 | ||
1255e269 DB |
2809 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, |
2810 | unsigned int, vlen, unsigned int, flags, | |
c2e6c856 | 2811 | struct __kernel_timespec __user *, timeout) |
1255e269 | 2812 | { |
e11d4284 AB |
2813 | if (flags & MSG_CMSG_COMPAT) |
2814 | return -EINVAL; | |
2815 | ||
2816 | return __sys_recvmmsg(fd, mmsg, vlen, flags, timeout, NULL); | |
2817 | } | |
2818 | ||
2819 | #ifdef CONFIG_COMPAT_32BIT_TIME | |
2820 | SYSCALL_DEFINE5(recvmmsg_time32, int, fd, struct mmsghdr __user *, mmsg, | |
2821 | unsigned int, vlen, unsigned int, flags, | |
2822 | struct old_timespec32 __user *, timeout) | |
2823 | { | |
2824 | if (flags & MSG_CMSG_COMPAT) | |
2825 | return -EINVAL; | |
2826 | ||
2827 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL, timeout); | |
1255e269 | 2828 | } |
e11d4284 | 2829 | #endif |
1255e269 | 2830 | |
a2e27255 | 2831 | #ifdef __ARCH_WANT_SYS_SOCKETCALL |
1da177e4 LT |
2832 | /* Argument list sizes for sys_socketcall */ |
2833 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2834 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2835 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2836 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2837 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2838 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2839 | }; |
2840 | ||
1da177e4 LT |
2841 | #undef AL |
2842 | ||
2843 | /* | |
89bddce5 | 2844 | * System call vectors. |
1da177e4 LT |
2845 | * |
2846 | * Argument checking cleaned up. Saved 20% in size. | |
2847 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2848 | * it is set by the callees. |
1da177e4 LT |
2849 | */ |
2850 | ||
3e0fa65f | 2851 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2852 | { |
2950fa9d | 2853 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2854 | unsigned long a0, a1; |
1da177e4 | 2855 | int err; |
47379052 | 2856 | unsigned int len; |
1da177e4 | 2857 | |
228e548e | 2858 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 | 2859 | return -EINVAL; |
c8e8cd57 | 2860 | call = array_index_nospec(call, SYS_SENDMMSG + 1); |
1da177e4 | 2861 | |
47379052 AV |
2862 | len = nargs[call]; |
2863 | if (len > sizeof(a)) | |
2864 | return -EINVAL; | |
2865 | ||
1da177e4 | 2866 | /* copy_from_user should be SMP safe. */ |
47379052 | 2867 | if (copy_from_user(a, args, len)) |
1da177e4 | 2868 | return -EFAULT; |
3ec3b2fb | 2869 | |
2950fa9d CG |
2870 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2871 | if (err) | |
2872 | return err; | |
3ec3b2fb | 2873 | |
89bddce5 SH |
2874 | a0 = a[0]; |
2875 | a1 = a[1]; | |
2876 | ||
2877 | switch (call) { | |
2878 | case SYS_SOCKET: | |
9d6a15c3 | 2879 | err = __sys_socket(a0, a1, a[2]); |
89bddce5 SH |
2880 | break; |
2881 | case SYS_BIND: | |
a87d35d8 | 2882 | err = __sys_bind(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2883 | break; |
2884 | case SYS_CONNECT: | |
1387c2c2 | 2885 | err = __sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2886 | break; |
2887 | case SYS_LISTEN: | |
25e290ee | 2888 | err = __sys_listen(a0, a1); |
89bddce5 SH |
2889 | break; |
2890 | case SYS_ACCEPT: | |
4541e805 DB |
2891 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2892 | (int __user *)a[2], 0); | |
89bddce5 SH |
2893 | break; |
2894 | case SYS_GETSOCKNAME: | |
2895 | err = | |
8882a107 DB |
2896 | __sys_getsockname(a0, (struct sockaddr __user *)a1, |
2897 | (int __user *)a[2]); | |
89bddce5 SH |
2898 | break; |
2899 | case SYS_GETPEERNAME: | |
2900 | err = | |
b21c8f83 DB |
2901 | __sys_getpeername(a0, (struct sockaddr __user *)a1, |
2902 | (int __user *)a[2]); | |
89bddce5 SH |
2903 | break; |
2904 | case SYS_SOCKETPAIR: | |
6debc8d8 | 2905 | err = __sys_socketpair(a0, a1, a[2], (int __user *)a[3]); |
89bddce5 SH |
2906 | break; |
2907 | case SYS_SEND: | |
f3bf896b DB |
2908 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2909 | NULL, 0); | |
89bddce5 SH |
2910 | break; |
2911 | case SYS_SENDTO: | |
211b634b DB |
2912 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2913 | (struct sockaddr __user *)a[4], a[5]); | |
89bddce5 SH |
2914 | break; |
2915 | case SYS_RECV: | |
d27e9afc DB |
2916 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2917 | NULL, NULL); | |
89bddce5 SH |
2918 | break; |
2919 | case SYS_RECVFROM: | |
7a09e1eb DB |
2920 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2921 | (struct sockaddr __user *)a[4], | |
2922 | (int __user *)a[5]); | |
89bddce5 SH |
2923 | break; |
2924 | case SYS_SHUTDOWN: | |
005a1aea | 2925 | err = __sys_shutdown(a0, a1); |
89bddce5 SH |
2926 | break; |
2927 | case SYS_SETSOCKOPT: | |
cc36dca0 DB |
2928 | err = __sys_setsockopt(a0, a1, a[2], (char __user *)a[3], |
2929 | a[4]); | |
89bddce5 SH |
2930 | break; |
2931 | case SYS_GETSOCKOPT: | |
2932 | err = | |
13a2d70e DB |
2933 | __sys_getsockopt(a0, a1, a[2], (char __user *)a[3], |
2934 | (int __user *)a[4]); | |
89bddce5 SH |
2935 | break; |
2936 | case SYS_SENDMSG: | |
e1834a32 DB |
2937 | err = __sys_sendmsg(a0, (struct user_msghdr __user *)a1, |
2938 | a[2], true); | |
89bddce5 | 2939 | break; |
228e548e | 2940 | case SYS_SENDMMSG: |
e1834a32 DB |
2941 | err = __sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
2942 | a[3], true); | |
228e548e | 2943 | break; |
89bddce5 | 2944 | case SYS_RECVMSG: |
e1834a32 DB |
2945 | err = __sys_recvmsg(a0, (struct user_msghdr __user *)a1, |
2946 | a[2], true); | |
89bddce5 | 2947 | break; |
a2e27255 | 2948 | case SYS_RECVMMSG: |
3ca47e95 | 2949 | if (IS_ENABLED(CONFIG_64BIT)) |
e11d4284 AB |
2950 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, |
2951 | a[2], a[3], | |
2952 | (struct __kernel_timespec __user *)a[4], | |
2953 | NULL); | |
2954 | else | |
2955 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, | |
2956 | a[2], a[3], NULL, | |
2957 | (struct old_timespec32 __user *)a[4]); | |
a2e27255 | 2958 | break; |
de11defe | 2959 | case SYS_ACCEPT4: |
4541e805 DB |
2960 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2961 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2962 | break; |
89bddce5 SH |
2963 | default: |
2964 | err = -EINVAL; | |
2965 | break; | |
1da177e4 LT |
2966 | } |
2967 | return err; | |
2968 | } | |
2969 | ||
89bddce5 | 2970 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2971 | |
55737fda SH |
2972 | /** |
2973 | * sock_register - add a socket protocol handler | |
2974 | * @ops: description of protocol | |
2975 | * | |
1da177e4 LT |
2976 | * This function is called by a protocol handler that wants to |
2977 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2978 | * socket interface. The value ops->family corresponds to the |
55737fda | 2979 | * socket system call protocol family. |
1da177e4 | 2980 | */ |
f0fd27d4 | 2981 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2982 | { |
2983 | int err; | |
2984 | ||
2985 | if (ops->family >= NPROTO) { | |
3410f22e | 2986 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2987 | return -ENOBUFS; |
2988 | } | |
55737fda SH |
2989 | |
2990 | spin_lock(&net_family_lock); | |
190683a9 ED |
2991 | if (rcu_dereference_protected(net_families[ops->family], |
2992 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2993 | err = -EEXIST; |
2994 | else { | |
cf778b00 | 2995 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2996 | err = 0; |
2997 | } | |
55737fda SH |
2998 | spin_unlock(&net_family_lock); |
2999 | ||
3410f22e | 3000 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
3001 | return err; |
3002 | } | |
c6d409cf | 3003 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 3004 | |
55737fda SH |
3005 | /** |
3006 | * sock_unregister - remove a protocol handler | |
3007 | * @family: protocol family to remove | |
3008 | * | |
1da177e4 LT |
3009 | * This function is called by a protocol handler that wants to |
3010 | * remove its address family, and have it unlinked from the | |
55737fda SH |
3011 | * new socket creation. |
3012 | * | |
3013 | * If protocol handler is a module, then it can use module reference | |
3014 | * counts to protect against new references. If protocol handler is not | |
3015 | * a module then it needs to provide its own protection in | |
3016 | * the ops->create routine. | |
1da177e4 | 3017 | */ |
f0fd27d4 | 3018 | void sock_unregister(int family) |
1da177e4 | 3019 | { |
f0fd27d4 | 3020 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 3021 | |
55737fda | 3022 | spin_lock(&net_family_lock); |
a9b3cd7f | 3023 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
3024 | spin_unlock(&net_family_lock); |
3025 | ||
3026 | synchronize_rcu(); | |
3027 | ||
3410f22e | 3028 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 3029 | } |
c6d409cf | 3030 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 3031 | |
bf2ae2e4 XL |
3032 | bool sock_is_registered(int family) |
3033 | { | |
66b51b0a | 3034 | return family < NPROTO && rcu_access_pointer(net_families[family]); |
bf2ae2e4 XL |
3035 | } |
3036 | ||
77d76ea3 | 3037 | static int __init sock_init(void) |
1da177e4 | 3038 | { |
b3e19d92 | 3039 | int err; |
2ca794e5 EB |
3040 | /* |
3041 | * Initialize the network sysctl infrastructure. | |
3042 | */ | |
3043 | err = net_sysctl_init(); | |
3044 | if (err) | |
3045 | goto out; | |
b3e19d92 | 3046 | |
1da177e4 | 3047 | /* |
89bddce5 | 3048 | * Initialize skbuff SLAB cache |
1da177e4 LT |
3049 | */ |
3050 | skb_init(); | |
1da177e4 LT |
3051 | |
3052 | /* | |
89bddce5 | 3053 | * Initialize the protocols module. |
1da177e4 LT |
3054 | */ |
3055 | ||
3056 | init_inodecache(); | |
b3e19d92 NP |
3057 | |
3058 | err = register_filesystem(&sock_fs_type); | |
3059 | if (err) | |
3060 | goto out_fs; | |
1da177e4 | 3061 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
3062 | if (IS_ERR(sock_mnt)) { |
3063 | err = PTR_ERR(sock_mnt); | |
3064 | goto out_mount; | |
3065 | } | |
77d76ea3 AK |
3066 | |
3067 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
3068 | */ |
3069 | ||
3070 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
3071 | err = netfilter_init(); |
3072 | if (err) | |
3073 | goto out; | |
1da177e4 | 3074 | #endif |
cbeb321a | 3075 | |
408eccce | 3076 | ptp_classifier_init(); |
c1f19b51 | 3077 | |
b3e19d92 NP |
3078 | out: |
3079 | return err; | |
3080 | ||
3081 | out_mount: | |
3082 | unregister_filesystem(&sock_fs_type); | |
3083 | out_fs: | |
3084 | goto out; | |
1da177e4 LT |
3085 | } |
3086 | ||
77d76ea3 AK |
3087 | core_initcall(sock_init); /* early initcall */ |
3088 | ||
1da177e4 LT |
3089 | #ifdef CONFIG_PROC_FS |
3090 | void socket_seq_show(struct seq_file *seq) | |
3091 | { | |
648845ab TZ |
3092 | seq_printf(seq, "sockets: used %d\n", |
3093 | sock_inuse_get(seq->private)); | |
1da177e4 | 3094 | } |
89bddce5 | 3095 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 3096 | |
89bbfc95 | 3097 | #ifdef CONFIG_COMPAT |
36fd633e | 3098 | static int compat_dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 3099 | { |
6b96018b | 3100 | struct compat_ifconf ifc32; |
7a229387 | 3101 | struct ifconf ifc; |
7a229387 AB |
3102 | int err; |
3103 | ||
6b96018b | 3104 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
3105 | return -EFAULT; |
3106 | ||
36fd633e AV |
3107 | ifc.ifc_len = ifc32.ifc_len; |
3108 | ifc.ifc_req = compat_ptr(ifc32.ifcbuf); | |
7a229387 | 3109 | |
36fd633e AV |
3110 | rtnl_lock(); |
3111 | err = dev_ifconf(net, &ifc, sizeof(struct compat_ifreq)); | |
3112 | rtnl_unlock(); | |
7a229387 AB |
3113 | if (err) |
3114 | return err; | |
3115 | ||
36fd633e | 3116 | ifc32.ifc_len = ifc.ifc_len; |
6b96018b | 3117 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
3118 | return -EFAULT; |
3119 | ||
3120 | return 0; | |
3121 | } | |
3122 | ||
6b96018b | 3123 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 3124 | { |
3a7da39d BH |
3125 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
3126 | bool convert_in = false, convert_out = false; | |
44c02a2c AV |
3127 | size_t buf_size = 0; |
3128 | struct ethtool_rxnfc __user *rxnfc = NULL; | |
3129 | struct ifreq ifr; | |
3a7da39d BH |
3130 | u32 rule_cnt = 0, actual_rule_cnt; |
3131 | u32 ethcmd; | |
7a229387 | 3132 | u32 data; |
3a7da39d | 3133 | int ret; |
7a229387 | 3134 | |
3a7da39d BH |
3135 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
3136 | return -EFAULT; | |
7a229387 | 3137 | |
3a7da39d BH |
3138 | compat_rxnfc = compat_ptr(data); |
3139 | ||
3140 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
3141 | return -EFAULT; |
3142 | ||
3a7da39d BH |
3143 | /* Most ethtool structures are defined without padding. |
3144 | * Unfortunately struct ethtool_rxnfc is an exception. | |
3145 | */ | |
3146 | switch (ethcmd) { | |
3147 | default: | |
3148 | break; | |
3149 | case ETHTOOL_GRXCLSRLALL: | |
3150 | /* Buffer size is variable */ | |
3151 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
3152 | return -EFAULT; | |
3153 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
3154 | return -ENOMEM; | |
3155 | buf_size += rule_cnt * sizeof(u32); | |
3156 | /* fall through */ | |
3157 | case ETHTOOL_GRXRINGS: | |
3158 | case ETHTOOL_GRXCLSRLCNT: | |
3159 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 3160 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
3161 | convert_out = true; |
3162 | /* fall through */ | |
3163 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
3164 | buf_size += sizeof(struct ethtool_rxnfc); |
3165 | convert_in = true; | |
44c02a2c | 3166 | rxnfc = compat_alloc_user_space(buf_size); |
3a7da39d BH |
3167 | break; |
3168 | } | |
3169 | ||
44c02a2c | 3170 | if (copy_from_user(&ifr.ifr_name, &ifr32->ifr_name, IFNAMSIZ)) |
7a229387 AB |
3171 | return -EFAULT; |
3172 | ||
44c02a2c | 3173 | ifr.ifr_data = convert_in ? rxnfc : (void __user *)compat_rxnfc; |
7a229387 | 3174 | |
3a7da39d | 3175 | if (convert_in) { |
127fe533 | 3176 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
3177 | * fs.ring_cookie and at the end of fs, but nowhere else. |
3178 | */ | |
127fe533 AD |
3179 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
3180 | sizeof(compat_rxnfc->fs.m_ext) != | |
3181 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
3182 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
3183 | BUILD_BUG_ON( |
3184 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
3185 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
3186 | offsetof(struct ethtool_rxnfc, fs.location) - | |
3187 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
3188 | ||
3189 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
3190 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
3191 | (void __user *)rxnfc) || | |
3a7da39d BH |
3192 | copy_in_user(&rxnfc->fs.ring_cookie, |
3193 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 | 3194 | (void __user *)(&rxnfc->fs.location + 1) - |
b6168562 WW |
3195 | (void __user *)&rxnfc->fs.ring_cookie)) |
3196 | return -EFAULT; | |
3197 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3198 | if (put_user(rule_cnt, &rxnfc->rule_cnt)) | |
3199 | return -EFAULT; | |
3200 | } else if (copy_in_user(&rxnfc->rule_cnt, | |
3201 | &compat_rxnfc->rule_cnt, | |
3202 | sizeof(rxnfc->rule_cnt))) | |
3a7da39d BH |
3203 | return -EFAULT; |
3204 | } | |
3205 | ||
44c02a2c | 3206 | ret = dev_ioctl(net, SIOCETHTOOL, &ifr, NULL); |
3a7da39d BH |
3207 | if (ret) |
3208 | return ret; | |
3209 | ||
3210 | if (convert_out) { | |
3211 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
3212 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
3213 | (const void __user *)rxnfc) || | |
3a7da39d BH |
3214 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
3215 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
3216 | (const void __user *)(&rxnfc->fs.location + 1) - |
3217 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
3218 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
3219 | sizeof(rxnfc->rule_cnt))) | |
3220 | return -EFAULT; | |
3221 | ||
3222 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3223 | /* As an optimisation, we only copy the actual | |
3224 | * number of rules that the underlying | |
3225 | * function returned. Since Mallory might | |
3226 | * change the rule count in user memory, we | |
3227 | * check that it is less than the rule count | |
3228 | * originally given (as the user buffer size), | |
3229 | * which has been range-checked. | |
3230 | */ | |
3231 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
3232 | return -EFAULT; | |
3233 | if (actual_rule_cnt < rule_cnt) | |
3234 | rule_cnt = actual_rule_cnt; | |
3235 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
3236 | &rxnfc->rule_locs[0], | |
3237 | rule_cnt * sizeof(u32))) | |
3238 | return -EFAULT; | |
3239 | } | |
3240 | } | |
3241 | ||
3242 | return 0; | |
7a229387 AB |
3243 | } |
3244 | ||
7a50a240 AB |
3245 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
3246 | { | |
7a50a240 | 3247 | compat_uptr_t uptr32; |
44c02a2c AV |
3248 | struct ifreq ifr; |
3249 | void __user *saved; | |
3250 | int err; | |
7a50a240 | 3251 | |
44c02a2c | 3252 | if (copy_from_user(&ifr, uifr32, sizeof(struct compat_ifreq))) |
7a50a240 AB |
3253 | return -EFAULT; |
3254 | ||
3255 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
3256 | return -EFAULT; | |
3257 | ||
44c02a2c AV |
3258 | saved = ifr.ifr_settings.ifs_ifsu.raw_hdlc; |
3259 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = compat_ptr(uptr32); | |
7a229387 | 3260 | |
44c02a2c AV |
3261 | err = dev_ioctl(net, SIOCWANDEV, &ifr, NULL); |
3262 | if (!err) { | |
3263 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = saved; | |
3264 | if (copy_to_user(uifr32, &ifr, sizeof(struct compat_ifreq))) | |
3265 | err = -EFAULT; | |
ccbd6a5a | 3266 | } |
44c02a2c | 3267 | return err; |
7a229387 AB |
3268 | } |
3269 | ||
590d4693 BH |
3270 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
3271 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 3272 | struct compat_ifreq __user *u_ifreq32) |
7a229387 | 3273 | { |
44c02a2c | 3274 | struct ifreq ifreq; |
7a229387 AB |
3275 | u32 data32; |
3276 | ||
44c02a2c | 3277 | if (copy_from_user(ifreq.ifr_name, u_ifreq32->ifr_name, IFNAMSIZ)) |
7a229387 | 3278 | return -EFAULT; |
44c02a2c | 3279 | if (get_user(data32, &u_ifreq32->ifr_data)) |
7a229387 | 3280 | return -EFAULT; |
44c02a2c | 3281 | ifreq.ifr_data = compat_ptr(data32); |
7a229387 | 3282 | |
44c02a2c | 3283 | return dev_ioctl(net, cmd, &ifreq, NULL); |
7a229387 AB |
3284 | } |
3285 | ||
37ac39bd JB |
3286 | static int compat_ifreq_ioctl(struct net *net, struct socket *sock, |
3287 | unsigned int cmd, | |
3288 | struct compat_ifreq __user *uifr32) | |
3289 | { | |
3290 | struct ifreq __user *uifr; | |
3291 | int err; | |
3292 | ||
3293 | /* Handle the fact that while struct ifreq has the same *layout* on | |
3294 | * 32/64 for everything but ifreq::ifru_ifmap and ifreq::ifru_data, | |
3295 | * which are handled elsewhere, it still has different *size* due to | |
3296 | * ifreq::ifru_ifmap (which is 16 bytes on 32 bit, 24 bytes on 64-bit, | |
3297 | * resulting in struct ifreq being 32 and 40 bytes respectively). | |
3298 | * As a result, if the struct happens to be at the end of a page and | |
3299 | * the next page isn't readable/writable, we get a fault. To prevent | |
3300 | * that, copy back and forth to the full size. | |
3301 | */ | |
3302 | ||
3303 | uifr = compat_alloc_user_space(sizeof(*uifr)); | |
3304 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
3305 | return -EFAULT; | |
3306 | ||
3307 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
3308 | ||
3309 | if (!err) { | |
3310 | switch (cmd) { | |
3311 | case SIOCGIFFLAGS: | |
3312 | case SIOCGIFMETRIC: | |
3313 | case SIOCGIFMTU: | |
3314 | case SIOCGIFMEM: | |
3315 | case SIOCGIFHWADDR: | |
3316 | case SIOCGIFINDEX: | |
3317 | case SIOCGIFADDR: | |
3318 | case SIOCGIFBRDADDR: | |
3319 | case SIOCGIFDSTADDR: | |
3320 | case SIOCGIFNETMASK: | |
3321 | case SIOCGIFPFLAGS: | |
3322 | case SIOCGIFTXQLEN: | |
3323 | case SIOCGMIIPHY: | |
3324 | case SIOCGMIIREG: | |
c6c9fee3 | 3325 | case SIOCGIFNAME: |
37ac39bd JB |
3326 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
3327 | err = -EFAULT; | |
3328 | break; | |
3329 | } | |
3330 | } | |
3331 | return err; | |
3332 | } | |
3333 | ||
a2116ed2 AB |
3334 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
3335 | struct compat_ifreq __user *uifr32) | |
3336 | { | |
3337 | struct ifreq ifr; | |
3338 | struct compat_ifmap __user *uifmap32; | |
a2116ed2 AB |
3339 | int err; |
3340 | ||
3341 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
3342 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3343 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3344 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3345 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3346 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3347 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3348 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3349 | if (err) |
3350 | return -EFAULT; | |
3351 | ||
44c02a2c | 3352 | err = dev_ioctl(net, cmd, &ifr, NULL); |
a2116ed2 AB |
3353 | |
3354 | if (cmd == SIOCGIFMAP && !err) { | |
3355 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3356 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3357 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3358 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3359 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3360 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3361 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3362 | if (err) |
3363 | err = -EFAULT; | |
3364 | } | |
3365 | return err; | |
3366 | } | |
3367 | ||
7a229387 | 3368 | struct rtentry32 { |
c6d409cf | 3369 | u32 rt_pad1; |
7a229387 AB |
3370 | struct sockaddr rt_dst; /* target address */ |
3371 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
3372 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
3373 | unsigned short rt_flags; |
3374 | short rt_pad2; | |
3375 | u32 rt_pad3; | |
3376 | unsigned char rt_tos; | |
3377 | unsigned char rt_class; | |
3378 | short rt_pad4; | |
3379 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 3380 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
3381 | u32 rt_mtu; /* per route MTU/Window */ |
3382 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
3383 | unsigned short rt_irtt; /* Initial RTT */ |
3384 | }; | |
3385 | ||
3386 | struct in6_rtmsg32 { | |
3387 | struct in6_addr rtmsg_dst; | |
3388 | struct in6_addr rtmsg_src; | |
3389 | struct in6_addr rtmsg_gateway; | |
3390 | u32 rtmsg_type; | |
3391 | u16 rtmsg_dst_len; | |
3392 | u16 rtmsg_src_len; | |
3393 | u32 rtmsg_metric; | |
3394 | u32 rtmsg_info; | |
3395 | u32 rtmsg_flags; | |
3396 | s32 rtmsg_ifindex; | |
3397 | }; | |
3398 | ||
6b96018b AB |
3399 | static int routing_ioctl(struct net *net, struct socket *sock, |
3400 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3401 | { |
3402 | int ret; | |
3403 | void *r = NULL; | |
3404 | struct in6_rtmsg r6; | |
3405 | struct rtentry r4; | |
3406 | char devname[16]; | |
3407 | u32 rtdev; | |
3408 | mm_segment_t old_fs = get_fs(); | |
3409 | ||
6b96018b AB |
3410 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3411 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3412 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3413 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
3414 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3415 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3416 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3417 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3418 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3419 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3420 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3421 | |
3422 | r = (void *) &r6; | |
3423 | } else { /* ipv4 */ | |
6b96018b | 3424 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3425 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3426 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
3427 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
3428 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3429 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3430 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
3431 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3432 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3433 | if (rtdev) { |
c6d409cf | 3434 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3435 | r4.rt_dev = (char __user __force *)devname; |
3436 | devname[15] = 0; | |
7a229387 AB |
3437 | } else |
3438 | r4.rt_dev = NULL; | |
3439 | ||
3440 | r = (void *) &r4; | |
3441 | } | |
3442 | ||
3443 | if (ret) { | |
3444 | ret = -EFAULT; | |
3445 | goto out; | |
3446 | } | |
3447 | ||
c6d409cf | 3448 | set_fs(KERNEL_DS); |
63ff03ab | 3449 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3450 | set_fs(old_fs); |
7a229387 AB |
3451 | |
3452 | out: | |
7a229387 AB |
3453 | return ret; |
3454 | } | |
3455 | ||
3456 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3457 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3458 | * use compatible ioctls |
7a229387 | 3459 | */ |
6b96018b | 3460 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3461 | { |
6b96018b | 3462 | compat_ulong_t tmp; |
7a229387 | 3463 | |
6b96018b | 3464 | if (get_user(tmp, argp)) |
7a229387 AB |
3465 | return -EFAULT; |
3466 | if (tmp == BRCTL_GET_VERSION) | |
3467 | return BRCTL_VERSION + 1; | |
3468 | return -EINVAL; | |
3469 | } | |
3470 | ||
6b96018b AB |
3471 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3472 | unsigned int cmd, unsigned long arg) | |
3473 | { | |
3474 | void __user *argp = compat_ptr(arg); | |
3475 | struct sock *sk = sock->sk; | |
3476 | struct net *net = sock_net(sk); | |
7a229387 | 3477 | |
6b96018b | 3478 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3479 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3480 | |
3481 | switch (cmd) { | |
3482 | case SIOCSIFBR: | |
3483 | case SIOCGIFBR: | |
3484 | return old_bridge_ioctl(argp); | |
6b96018b | 3485 | case SIOCGIFCONF: |
36fd633e | 3486 | return compat_dev_ifconf(net, argp); |
6b96018b AB |
3487 | case SIOCETHTOOL: |
3488 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3489 | case SIOCWANDEV: |
3490 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3491 | case SIOCGIFMAP: |
3492 | case SIOCSIFMAP: | |
3493 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3494 | case SIOCADDRT: |
3495 | case SIOCDELRT: | |
3496 | return routing_ioctl(net, sock, cmd, argp); | |
0768e170 AB |
3497 | case SIOCGSTAMP_OLD: |
3498 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
3499 | if (!sock->ops->gettstamp) |
3500 | return -ENOIOCTLCMD; | |
0768e170 | 3501 | return sock->ops->gettstamp(sock, argp, cmd == SIOCGSTAMP_OLD, |
c7cbdbf2 AB |
3502 | !COMPAT_USE_64BIT_TIME); |
3503 | ||
590d4693 BH |
3504 | case SIOCBONDSLAVEINFOQUERY: |
3505 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3506 | case SIOCSHWTSTAMP: |
fd468c74 | 3507 | case SIOCGHWTSTAMP: |
590d4693 | 3508 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3509 | |
3510 | case FIOSETOWN: | |
3511 | case SIOCSPGRP: | |
3512 | case FIOGETOWN: | |
3513 | case SIOCGPGRP: | |
3514 | case SIOCBRADDBR: | |
3515 | case SIOCBRDELBR: | |
3516 | case SIOCGIFVLAN: | |
3517 | case SIOCSIFVLAN: | |
3518 | case SIOCADDDLCI: | |
3519 | case SIOCDELDLCI: | |
c62cce2c | 3520 | case SIOCGSKNS: |
0768e170 AB |
3521 | case SIOCGSTAMP_NEW: |
3522 | case SIOCGSTAMPNS_NEW: | |
6b96018b AB |
3523 | return sock_ioctl(file, cmd, arg); |
3524 | ||
3525 | case SIOCGIFFLAGS: | |
3526 | case SIOCSIFFLAGS: | |
3527 | case SIOCGIFMETRIC: | |
3528 | case SIOCSIFMETRIC: | |
3529 | case SIOCGIFMTU: | |
3530 | case SIOCSIFMTU: | |
3531 | case SIOCGIFMEM: | |
3532 | case SIOCSIFMEM: | |
3533 | case SIOCGIFHWADDR: | |
3534 | case SIOCSIFHWADDR: | |
3535 | case SIOCADDMULTI: | |
3536 | case SIOCDELMULTI: | |
3537 | case SIOCGIFINDEX: | |
6b96018b AB |
3538 | case SIOCGIFADDR: |
3539 | case SIOCSIFADDR: | |
3540 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3541 | case SIOCDIFADDR: |
6b96018b AB |
3542 | case SIOCGIFBRDADDR: |
3543 | case SIOCSIFBRDADDR: | |
3544 | case SIOCGIFDSTADDR: | |
3545 | case SIOCSIFDSTADDR: | |
3546 | case SIOCGIFNETMASK: | |
3547 | case SIOCSIFNETMASK: | |
3548 | case SIOCSIFPFLAGS: | |
3549 | case SIOCGIFPFLAGS: | |
3550 | case SIOCGIFTXQLEN: | |
3551 | case SIOCSIFTXQLEN: | |
3552 | case SIOCBRADDIF: | |
3553 | case SIOCBRDELIF: | |
c6c9fee3 | 3554 | case SIOCGIFNAME: |
9177efd3 AB |
3555 | case SIOCSIFNAME: |
3556 | case SIOCGMIIPHY: | |
3557 | case SIOCGMIIREG: | |
3558 | case SIOCSMIIREG: | |
f92d4fc9 AV |
3559 | case SIOCBONDENSLAVE: |
3560 | case SIOCBONDRELEASE: | |
3561 | case SIOCBONDSETHWADDR: | |
3562 | case SIOCBONDCHANGEACTIVE: | |
37ac39bd JB |
3563 | return compat_ifreq_ioctl(net, sock, cmd, argp); |
3564 | ||
6b96018b AB |
3565 | case SIOCSARP: |
3566 | case SIOCGARP: | |
3567 | case SIOCDARP: | |
c7dc504e | 3568 | case SIOCOUTQ: |
9d7bf41f | 3569 | case SIOCOUTQNSD: |
6b96018b | 3570 | case SIOCATMARK: |
63ff03ab | 3571 | return sock_do_ioctl(net, sock, cmd, arg); |
9177efd3 AB |
3572 | } |
3573 | ||
6b96018b AB |
3574 | return -ENOIOCTLCMD; |
3575 | } | |
7a229387 | 3576 | |
95c96174 | 3577 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3578 | unsigned long arg) |
89bbfc95 SP |
3579 | { |
3580 | struct socket *sock = file->private_data; | |
3581 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3582 | struct sock *sk; |
3583 | struct net *net; | |
3584 | ||
3585 | sk = sock->sk; | |
3586 | net = sock_net(sk); | |
89bbfc95 SP |
3587 | |
3588 | if (sock->ops->compat_ioctl) | |
3589 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3590 | ||
87de87d5 DM |
3591 | if (ret == -ENOIOCTLCMD && |
3592 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3593 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3594 | ||
6b96018b AB |
3595 | if (ret == -ENOIOCTLCMD) |
3596 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3597 | ||
89bbfc95 SP |
3598 | return ret; |
3599 | } | |
3600 | #endif | |
3601 | ||
8a3c245c PT |
3602 | /** |
3603 | * kernel_bind - bind an address to a socket (kernel space) | |
3604 | * @sock: socket | |
3605 | * @addr: address | |
3606 | * @addrlen: length of address | |
3607 | * | |
3608 | * Returns 0 or an error. | |
3609 | */ | |
3610 | ||
ac5a488e SS |
3611 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3612 | { | |
3613 | return sock->ops->bind(sock, addr, addrlen); | |
3614 | } | |
c6d409cf | 3615 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e | 3616 | |
8a3c245c PT |
3617 | /** |
3618 | * kernel_listen - move socket to listening state (kernel space) | |
3619 | * @sock: socket | |
3620 | * @backlog: pending connections queue size | |
3621 | * | |
3622 | * Returns 0 or an error. | |
3623 | */ | |
3624 | ||
ac5a488e SS |
3625 | int kernel_listen(struct socket *sock, int backlog) |
3626 | { | |
3627 | return sock->ops->listen(sock, backlog); | |
3628 | } | |
c6d409cf | 3629 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e | 3630 | |
8a3c245c PT |
3631 | /** |
3632 | * kernel_accept - accept a connection (kernel space) | |
3633 | * @sock: listening socket | |
3634 | * @newsock: new connected socket | |
3635 | * @flags: flags | |
3636 | * | |
3637 | * @flags must be SOCK_CLOEXEC, SOCK_NONBLOCK or 0. | |
3638 | * If it fails, @newsock is guaranteed to be %NULL. | |
3639 | * Returns 0 or an error. | |
3640 | */ | |
3641 | ||
ac5a488e SS |
3642 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) |
3643 | { | |
3644 | struct sock *sk = sock->sk; | |
3645 | int err; | |
3646 | ||
3647 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3648 | newsock); | |
3649 | if (err < 0) | |
3650 | goto done; | |
3651 | ||
cdfbabfb | 3652 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3653 | if (err < 0) { |
3654 | sock_release(*newsock); | |
fa8705b0 | 3655 | *newsock = NULL; |
ac5a488e SS |
3656 | goto done; |
3657 | } | |
3658 | ||
3659 | (*newsock)->ops = sock->ops; | |
1b08534e | 3660 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3661 | |
3662 | done: | |
3663 | return err; | |
3664 | } | |
c6d409cf | 3665 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e | 3666 | |
8a3c245c PT |
3667 | /** |
3668 | * kernel_connect - connect a socket (kernel space) | |
3669 | * @sock: socket | |
3670 | * @addr: address | |
3671 | * @addrlen: address length | |
3672 | * @flags: flags (O_NONBLOCK, ...) | |
3673 | * | |
3674 | * For datagram sockets, @addr is the addres to which datagrams are sent | |
3675 | * by default, and the only address from which datagrams are received. | |
3676 | * For stream sockets, attempts to connect to @addr. | |
3677 | * Returns 0 or an error code. | |
3678 | */ | |
3679 | ||
ac5a488e | 3680 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, |
4768fbcb | 3681 | int flags) |
ac5a488e SS |
3682 | { |
3683 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3684 | } | |
c6d409cf | 3685 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e | 3686 | |
8a3c245c PT |
3687 | /** |
3688 | * kernel_getsockname - get the address which the socket is bound (kernel space) | |
3689 | * @sock: socket | |
3690 | * @addr: address holder | |
3691 | * | |
3692 | * Fills the @addr pointer with the address which the socket is bound. | |
3693 | * Returns 0 or an error code. | |
3694 | */ | |
3695 | ||
9b2c45d4 | 3696 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3697 | { |
9b2c45d4 | 3698 | return sock->ops->getname(sock, addr, 0); |
ac5a488e | 3699 | } |
c6d409cf | 3700 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e | 3701 | |
8a3c245c PT |
3702 | /** |
3703 | * kernel_peername - get the address which the socket is connected (kernel space) | |
3704 | * @sock: socket | |
3705 | * @addr: address holder | |
3706 | * | |
3707 | * Fills the @addr pointer with the address which the socket is connected. | |
3708 | * Returns 0 or an error code. | |
3709 | */ | |
3710 | ||
9b2c45d4 | 3711 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3712 | { |
9b2c45d4 | 3713 | return sock->ops->getname(sock, addr, 1); |
ac5a488e | 3714 | } |
c6d409cf | 3715 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e | 3716 | |
8a3c245c PT |
3717 | /** |
3718 | * kernel_getsockopt - get a socket option (kernel space) | |
3719 | * @sock: socket | |
3720 | * @level: API level (SOL_SOCKET, ...) | |
3721 | * @optname: option tag | |
3722 | * @optval: option value | |
3723 | * @optlen: option length | |
3724 | * | |
3725 | * Assigns the option length to @optlen. | |
3726 | * Returns 0 or an error. | |
3727 | */ | |
3728 | ||
ac5a488e SS |
3729 | int kernel_getsockopt(struct socket *sock, int level, int optname, |
3730 | char *optval, int *optlen) | |
3731 | { | |
3732 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3733 | char __user *uoptval; |
3734 | int __user *uoptlen; | |
ac5a488e SS |
3735 | int err; |
3736 | ||
fb8621bb NK |
3737 | uoptval = (char __user __force *) optval; |
3738 | uoptlen = (int __user __force *) optlen; | |
3739 | ||
ac5a488e SS |
3740 | set_fs(KERNEL_DS); |
3741 | if (level == SOL_SOCKET) | |
fb8621bb | 3742 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3743 | else |
fb8621bb NK |
3744 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3745 | uoptlen); | |
ac5a488e SS |
3746 | set_fs(oldfs); |
3747 | return err; | |
3748 | } | |
c6d409cf | 3749 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e | 3750 | |
8a3c245c PT |
3751 | /** |
3752 | * kernel_setsockopt - set a socket option (kernel space) | |
3753 | * @sock: socket | |
3754 | * @level: API level (SOL_SOCKET, ...) | |
3755 | * @optname: option tag | |
3756 | * @optval: option value | |
3757 | * @optlen: option length | |
3758 | * | |
3759 | * Returns 0 or an error. | |
3760 | */ | |
3761 | ||
ac5a488e | 3762 | int kernel_setsockopt(struct socket *sock, int level, int optname, |
b7058842 | 3763 | char *optval, unsigned int optlen) |
ac5a488e SS |
3764 | { |
3765 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3766 | char __user *uoptval; |
ac5a488e SS |
3767 | int err; |
3768 | ||
fb8621bb NK |
3769 | uoptval = (char __user __force *) optval; |
3770 | ||
ac5a488e SS |
3771 | set_fs(KERNEL_DS); |
3772 | if (level == SOL_SOCKET) | |
fb8621bb | 3773 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3774 | else |
fb8621bb | 3775 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3776 | optlen); |
3777 | set_fs(oldfs); | |
3778 | return err; | |
3779 | } | |
c6d409cf | 3780 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e | 3781 | |
8a3c245c PT |
3782 | /** |
3783 | * kernel_sendpage - send a &page through a socket (kernel space) | |
3784 | * @sock: socket | |
3785 | * @page: page | |
3786 | * @offset: page offset | |
3787 | * @size: total size in bytes | |
3788 | * @flags: flags (MSG_DONTWAIT, ...) | |
3789 | * | |
3790 | * Returns the total amount sent in bytes or an error. | |
3791 | */ | |
3792 | ||
ac5a488e SS |
3793 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, |
3794 | size_t size, int flags) | |
3795 | { | |
3796 | if (sock->ops->sendpage) | |
3797 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3798 | ||
3799 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3800 | } | |
c6d409cf | 3801 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3802 | |
8a3c245c PT |
3803 | /** |
3804 | * kernel_sendpage_locked - send a &page through the locked sock (kernel space) | |
3805 | * @sk: sock | |
3806 | * @page: page | |
3807 | * @offset: page offset | |
3808 | * @size: total size in bytes | |
3809 | * @flags: flags (MSG_DONTWAIT, ...) | |
3810 | * | |
3811 | * Returns the total amount sent in bytes or an error. | |
3812 | * Caller must hold @sk. | |
3813 | */ | |
3814 | ||
306b13eb TH |
3815 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3816 | size_t size, int flags) | |
3817 | { | |
3818 | struct socket *sock = sk->sk_socket; | |
3819 | ||
3820 | if (sock->ops->sendpage_locked) | |
3821 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3822 | flags); | |
3823 | ||
3824 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3825 | } | |
3826 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3827 | ||
8a3c245c PT |
3828 | /** |
3829 | * kernel_shutdown - shut down part of a full-duplex connection (kernel space) | |
3830 | * @sock: socket | |
3831 | * @how: connection part | |
3832 | * | |
3833 | * Returns 0 or an error. | |
3834 | */ | |
3835 | ||
91cf45f0 TM |
3836 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3837 | { | |
3838 | return sock->ops->shutdown(sock, how); | |
3839 | } | |
91cf45f0 | 3840 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 | 3841 | |
8a3c245c PT |
3842 | /** |
3843 | * kernel_sock_ip_overhead - returns the IP overhead imposed by a socket | |
3844 | * @sk: socket | |
3845 | * | |
3846 | * This routine returns the IP overhead imposed by a socket i.e. | |
3847 | * the length of the underlying IP header, depending on whether | |
3848 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
3849 | * on at the socket. Assumes that the caller has a lock on the socket. | |
113c3075 | 3850 | */ |
8a3c245c | 3851 | |
113c3075 P |
3852 | u32 kernel_sock_ip_overhead(struct sock *sk) |
3853 | { | |
3854 | struct inet_sock *inet; | |
3855 | struct ip_options_rcu *opt; | |
3856 | u32 overhead = 0; | |
113c3075 P |
3857 | #if IS_ENABLED(CONFIG_IPV6) |
3858 | struct ipv6_pinfo *np; | |
3859 | struct ipv6_txoptions *optv6 = NULL; | |
3860 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3861 | ||
3862 | if (!sk) | |
3863 | return overhead; | |
3864 | ||
113c3075 P |
3865 | switch (sk->sk_family) { |
3866 | case AF_INET: | |
3867 | inet = inet_sk(sk); | |
3868 | overhead += sizeof(struct iphdr); | |
3869 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3870 | sock_owned_by_user(sk)); |
113c3075 P |
3871 | if (opt) |
3872 | overhead += opt->opt.optlen; | |
3873 | return overhead; | |
3874 | #if IS_ENABLED(CONFIG_IPV6) | |
3875 | case AF_INET6: | |
3876 | np = inet6_sk(sk); | |
3877 | overhead += sizeof(struct ipv6hdr); | |
3878 | if (np) | |
3879 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3880 | sock_owned_by_user(sk)); |
113c3075 P |
3881 | if (optv6) |
3882 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3883 | return overhead; | |
3884 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3885 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3886 | return overhead; | |
3887 | } | |
3888 | } | |
3889 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |