]>
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); |
542d3065 AB |
131 | |
132 | #ifdef CONFIG_PROC_FS | |
133 | static void sock_show_fdinfo(struct seq_file *m, struct file *f) | |
134 | { | |
135 | struct socket *sock = f->private_data; | |
136 | ||
137 | if (sock->ops->show_fdinfo) | |
138 | sock->ops->show_fdinfo(m, sock); | |
139 | } | |
140 | #else | |
141 | #define sock_show_fdinfo NULL | |
142 | #endif | |
1da177e4 | 143 | |
1da177e4 LT |
144 | /* |
145 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
146 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
147 | */ | |
148 | ||
da7071d7 | 149 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
150 | .owner = THIS_MODULE, |
151 | .llseek = no_llseek, | |
8ae5e030 AV |
152 | .read_iter = sock_read_iter, |
153 | .write_iter = sock_write_iter, | |
1da177e4 LT |
154 | .poll = sock_poll, |
155 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
156 | #ifdef CONFIG_COMPAT |
157 | .compat_ioctl = compat_sock_ioctl, | |
158 | #endif | |
1da177e4 | 159 | .mmap = sock_mmap, |
1da177e4 LT |
160 | .release = sock_close, |
161 | .fasync = sock_fasync, | |
5274f052 JA |
162 | .sendpage = sock_sendpage, |
163 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 164 | .splice_read = sock_splice_read, |
b4653342 | 165 | .show_fdinfo = sock_show_fdinfo, |
1da177e4 LT |
166 | }; |
167 | ||
168 | /* | |
169 | * The protocol list. Each protocol is registered in here. | |
170 | */ | |
171 | ||
1da177e4 | 172 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 173 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 174 | |
1da177e4 | 175 | /* |
89bddce5 SH |
176 | * Support routines. |
177 | * Move socket addresses back and forth across the kernel/user | |
178 | * divide and look after the messy bits. | |
1da177e4 LT |
179 | */ |
180 | ||
1da177e4 LT |
181 | /** |
182 | * move_addr_to_kernel - copy a socket address into kernel space | |
183 | * @uaddr: Address in user space | |
184 | * @kaddr: Address in kernel space | |
185 | * @ulen: Length in user space | |
186 | * | |
187 | * The address is copied into kernel space. If the provided address is | |
188 | * too long an error code of -EINVAL is returned. If the copy gives | |
189 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
190 | */ | |
191 | ||
43db362d | 192 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 193 | { |
230b1839 | 194 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 195 | return -EINVAL; |
89bddce5 | 196 | if (ulen == 0) |
1da177e4 | 197 | return 0; |
89bddce5 | 198 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 199 | return -EFAULT; |
3ec3b2fb | 200 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
201 | } |
202 | ||
203 | /** | |
204 | * move_addr_to_user - copy an address to user space | |
205 | * @kaddr: kernel space address | |
206 | * @klen: length of address in kernel | |
207 | * @uaddr: user space address | |
208 | * @ulen: pointer to user length field | |
209 | * | |
210 | * The value pointed to by ulen on entry is the buffer length available. | |
211 | * This is overwritten with the buffer space used. -EINVAL is returned | |
212 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
213 | * is returned if either the buffer or the length field are not | |
214 | * accessible. | |
215 | * After copying the data up to the limit the user specifies, the true | |
216 | * length of the data is written over the length limit the user | |
217 | * specified. Zero is returned for a success. | |
218 | */ | |
89bddce5 | 219 | |
43db362d | 220 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 221 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
222 | { |
223 | int err; | |
224 | int len; | |
225 | ||
68c6beb3 | 226 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
227 | err = get_user(len, ulen); |
228 | if (err) | |
1da177e4 | 229 | return err; |
89bddce5 SH |
230 | if (len > klen) |
231 | len = klen; | |
68c6beb3 | 232 | if (len < 0) |
1da177e4 | 233 | return -EINVAL; |
89bddce5 | 234 | if (len) { |
d6fe3945 SG |
235 | if (audit_sockaddr(klen, kaddr)) |
236 | return -ENOMEM; | |
89bddce5 | 237 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
238 | return -EFAULT; |
239 | } | |
240 | /* | |
89bddce5 SH |
241 | * "fromlen shall refer to the value before truncation.." |
242 | * 1003.1g | |
1da177e4 LT |
243 | */ |
244 | return __put_user(klen, ulen); | |
245 | } | |
246 | ||
08009a76 | 247 | static struct kmem_cache *sock_inode_cachep __ro_after_init; |
1da177e4 LT |
248 | |
249 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
250 | { | |
251 | struct socket_alloc *ei; | |
89bddce5 | 252 | |
e94b1766 | 253 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
254 | if (!ei) |
255 | return NULL; | |
333f7909 AV |
256 | init_waitqueue_head(&ei->socket.wq.wait); |
257 | ei->socket.wq.fasync_list = NULL; | |
258 | ei->socket.wq.flags = 0; | |
89bddce5 | 259 | |
1da177e4 LT |
260 | ei->socket.state = SS_UNCONNECTED; |
261 | ei->socket.flags = 0; | |
262 | ei->socket.ops = NULL; | |
263 | ei->socket.sk = NULL; | |
264 | ei->socket.file = NULL; | |
1da177e4 LT |
265 | |
266 | return &ei->vfs_inode; | |
267 | } | |
268 | ||
6d7855c5 | 269 | static void sock_free_inode(struct inode *inode) |
1da177e4 | 270 | { |
43815482 ED |
271 | struct socket_alloc *ei; |
272 | ||
273 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
43815482 | 274 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
275 | } |
276 | ||
51cc5068 | 277 | static void init_once(void *foo) |
1da177e4 | 278 | { |
89bddce5 | 279 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 280 | |
a35afb83 | 281 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 282 | } |
89bddce5 | 283 | |
1e911632 | 284 | static void init_inodecache(void) |
1da177e4 LT |
285 | { |
286 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
287 | sizeof(struct socket_alloc), |
288 | 0, | |
289 | (SLAB_HWCACHE_ALIGN | | |
290 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 291 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 292 | init_once); |
1e911632 | 293 | BUG_ON(sock_inode_cachep == NULL); |
1da177e4 LT |
294 | } |
295 | ||
b87221de | 296 | static const struct super_operations sockfs_ops = { |
c6d409cf | 297 | .alloc_inode = sock_alloc_inode, |
6d7855c5 | 298 | .free_inode = sock_free_inode, |
c6d409cf | 299 | .statfs = simple_statfs, |
1da177e4 LT |
300 | }; |
301 | ||
c23fbb6b ED |
302 | /* |
303 | * sockfs_dname() is called from d_path(). | |
304 | */ | |
305 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
306 | { | |
307 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 308 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
309 | } |
310 | ||
3ba13d17 | 311 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 312 | .d_dname = sockfs_dname, |
1da177e4 LT |
313 | }; |
314 | ||
bba0bd31 AG |
315 | static int sockfs_xattr_get(const struct xattr_handler *handler, |
316 | struct dentry *dentry, struct inode *inode, | |
317 | const char *suffix, void *value, size_t size) | |
318 | { | |
319 | if (value) { | |
320 | if (dentry->d_name.len + 1 > size) | |
321 | return -ERANGE; | |
322 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
323 | } | |
324 | return dentry->d_name.len + 1; | |
325 | } | |
326 | ||
327 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" | |
328 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
329 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
330 | ||
331 | static const struct xattr_handler sockfs_xattr_handler = { | |
332 | .name = XATTR_NAME_SOCKPROTONAME, | |
333 | .get = sockfs_xattr_get, | |
334 | }; | |
335 | ||
4a590153 AG |
336 | static int sockfs_security_xattr_set(const struct xattr_handler *handler, |
337 | struct dentry *dentry, struct inode *inode, | |
338 | const char *suffix, const void *value, | |
339 | size_t size, int flags) | |
340 | { | |
341 | /* Handled by LSM. */ | |
342 | return -EAGAIN; | |
343 | } | |
344 | ||
345 | static const struct xattr_handler sockfs_security_xattr_handler = { | |
346 | .prefix = XATTR_SECURITY_PREFIX, | |
347 | .set = sockfs_security_xattr_set, | |
348 | }; | |
349 | ||
bba0bd31 AG |
350 | static const struct xattr_handler *sockfs_xattr_handlers[] = { |
351 | &sockfs_xattr_handler, | |
4a590153 | 352 | &sockfs_security_xattr_handler, |
bba0bd31 AG |
353 | NULL |
354 | }; | |
355 | ||
fba9be49 | 356 | static int sockfs_init_fs_context(struct fs_context *fc) |
c74a1cbb | 357 | { |
fba9be49 DH |
358 | struct pseudo_fs_context *ctx = init_pseudo(fc, SOCKFS_MAGIC); |
359 | if (!ctx) | |
360 | return -ENOMEM; | |
361 | ctx->ops = &sockfs_ops; | |
362 | ctx->dops = &sockfs_dentry_operations; | |
363 | ctx->xattr = sockfs_xattr_handlers; | |
364 | return 0; | |
c74a1cbb AV |
365 | } |
366 | ||
367 | static struct vfsmount *sock_mnt __read_mostly; | |
368 | ||
369 | static struct file_system_type sock_fs_type = { | |
370 | .name = "sockfs", | |
fba9be49 | 371 | .init_fs_context = sockfs_init_fs_context, |
c74a1cbb AV |
372 | .kill_sb = kill_anon_super, |
373 | }; | |
374 | ||
1da177e4 LT |
375 | /* |
376 | * Obtains the first available file descriptor and sets it up for use. | |
377 | * | |
39d8c1b6 DM |
378 | * These functions create file structures and maps them to fd space |
379 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
380 | * and file struct implicitly stored in sock->file. |
381 | * Note that another thread may close file descriptor before we return | |
382 | * from this function. We use the fact that now we do not refer | |
383 | * to socket after mapping. If one day we will need it, this | |
384 | * function will increment ref. count on file by 1. | |
385 | * | |
386 | * In any case returned fd MAY BE not valid! | |
387 | * This race condition is unavoidable | |
388 | * with shared fd spaces, we cannot solve it inside kernel, | |
389 | * but we take care of internal coherence yet. | |
390 | */ | |
391 | ||
8a3c245c PT |
392 | /** |
393 | * sock_alloc_file - Bind a &socket to a &file | |
394 | * @sock: socket | |
395 | * @flags: file status flags | |
396 | * @dname: protocol name | |
397 | * | |
398 | * Returns the &file bound with @sock, implicitly storing it | |
399 | * in sock->file. If dname is %NULL, sets to "". | |
400 | * On failure the return is a ERR pointer (see linux/err.h). | |
401 | * This function uses GFP_KERNEL internally. | |
402 | */ | |
403 | ||
aab174f0 | 404 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 405 | { |
7cbe66b6 | 406 | struct file *file; |
1da177e4 | 407 | |
d93aa9d8 AV |
408 | if (!dname) |
409 | dname = sock->sk ? sock->sk->sk_prot_creator->name : ""; | |
39d8c1b6 | 410 | |
d93aa9d8 AV |
411 | file = alloc_file_pseudo(SOCK_INODE(sock), sock_mnt, dname, |
412 | O_RDWR | (flags & O_NONBLOCK), | |
413 | &socket_file_ops); | |
b5ffe634 | 414 | if (IS_ERR(file)) { |
8e1611e2 | 415 | sock_release(sock); |
39b65252 | 416 | return file; |
cc3808f8 AV |
417 | } |
418 | ||
419 | sock->file = file; | |
39d8c1b6 | 420 | file->private_data = sock; |
d8e464ec | 421 | stream_open(SOCK_INODE(sock), file); |
28407630 | 422 | return file; |
39d8c1b6 | 423 | } |
56b31d1c | 424 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 425 | |
56b31d1c | 426 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
427 | { |
428 | struct file *newfile; | |
28407630 | 429 | int fd = get_unused_fd_flags(flags); |
ce4bb04c AV |
430 | if (unlikely(fd < 0)) { |
431 | sock_release(sock); | |
28407630 | 432 | return fd; |
ce4bb04c | 433 | } |
39d8c1b6 | 434 | |
aab174f0 | 435 | newfile = sock_alloc_file(sock, flags, NULL); |
4546e44c | 436 | if (!IS_ERR(newfile)) { |
39d8c1b6 | 437 | fd_install(fd, newfile); |
28407630 AV |
438 | return fd; |
439 | } | |
7cbe66b6 | 440 | |
28407630 AV |
441 | put_unused_fd(fd); |
442 | return PTR_ERR(newfile); | |
1da177e4 LT |
443 | } |
444 | ||
8a3c245c PT |
445 | /** |
446 | * sock_from_file - Return the &socket bounded to @file. | |
447 | * @file: file | |
448 | * @err: pointer to an error code return | |
449 | * | |
450 | * On failure returns %NULL and assigns -ENOTSOCK to @err. | |
451 | */ | |
452 | ||
406a3c63 | 453 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 454 | { |
6cb153ca BL |
455 | if (file->f_op == &socket_file_ops) |
456 | return file->private_data; /* set in sock_map_fd */ | |
457 | ||
23bb80d2 ED |
458 | *err = -ENOTSOCK; |
459 | return NULL; | |
6cb153ca | 460 | } |
406a3c63 | 461 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 462 | |
1da177e4 | 463 | /** |
c6d409cf | 464 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
465 | * @fd: file handle |
466 | * @err: pointer to an error code return | |
467 | * | |
468 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 469 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
470 | * with a negative errno code and NULL is returned. The function checks |
471 | * for both invalid handles and passing a handle which is not a socket. | |
472 | * | |
473 | * On a success the socket object pointer is returned. | |
474 | */ | |
475 | ||
476 | struct socket *sockfd_lookup(int fd, int *err) | |
477 | { | |
478 | struct file *file; | |
1da177e4 LT |
479 | struct socket *sock; |
480 | ||
89bddce5 SH |
481 | file = fget(fd); |
482 | if (!file) { | |
1da177e4 LT |
483 | *err = -EBADF; |
484 | return NULL; | |
485 | } | |
89bddce5 | 486 | |
6cb153ca BL |
487 | sock = sock_from_file(file, err); |
488 | if (!sock) | |
1da177e4 | 489 | fput(file); |
6cb153ca BL |
490 | return sock; |
491 | } | |
c6d409cf | 492 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 493 | |
6cb153ca BL |
494 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
495 | { | |
00e188ef | 496 | struct fd f = fdget(fd); |
6cb153ca BL |
497 | struct socket *sock; |
498 | ||
3672558c | 499 | *err = -EBADF; |
00e188ef AV |
500 | if (f.file) { |
501 | sock = sock_from_file(f.file, err); | |
502 | if (likely(sock)) { | |
503 | *fput_needed = f.flags; | |
6cb153ca | 504 | return sock; |
00e188ef AV |
505 | } |
506 | fdput(f); | |
1da177e4 | 507 | } |
6cb153ca | 508 | return NULL; |
1da177e4 LT |
509 | } |
510 | ||
600e1779 MY |
511 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
512 | size_t size) | |
513 | { | |
514 | ssize_t len; | |
515 | ssize_t used = 0; | |
516 | ||
c5ef6035 | 517 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
518 | if (len < 0) |
519 | return len; | |
520 | used += len; | |
521 | if (buffer) { | |
522 | if (size < used) | |
523 | return -ERANGE; | |
524 | buffer += len; | |
525 | } | |
526 | ||
527 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
528 | used += len; | |
529 | if (buffer) { | |
530 | if (size < used) | |
531 | return -ERANGE; | |
532 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
533 | buffer += len; | |
534 | } | |
535 | ||
536 | return used; | |
537 | } | |
538 | ||
dc647ec8 | 539 | static int sockfs_setattr(struct dentry *dentry, struct iattr *iattr) |
86741ec2 LC |
540 | { |
541 | int err = simple_setattr(dentry, iattr); | |
542 | ||
e1a3a60a | 543 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
544 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
545 | ||
6d8c50dc CW |
546 | if (sock->sk) |
547 | sock->sk->sk_uid = iattr->ia_uid; | |
548 | else | |
549 | err = -ENOENT; | |
86741ec2 LC |
550 | } |
551 | ||
552 | return err; | |
553 | } | |
554 | ||
600e1779 | 555 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 556 | .listxattr = sockfs_listxattr, |
86741ec2 | 557 | .setattr = sockfs_setattr, |
600e1779 MY |
558 | }; |
559 | ||
1da177e4 | 560 | /** |
8a3c245c | 561 | * sock_alloc - allocate a socket |
89bddce5 | 562 | * |
1da177e4 LT |
563 | * Allocate a new inode and socket object. The two are bound together |
564 | * and initialised. The socket is then returned. If we are out of inodes | |
8a3c245c | 565 | * NULL is returned. This functions uses GFP_KERNEL internally. |
1da177e4 LT |
566 | */ |
567 | ||
f4a00aac | 568 | struct socket *sock_alloc(void) |
1da177e4 | 569 | { |
89bddce5 SH |
570 | struct inode *inode; |
571 | struct socket *sock; | |
1da177e4 | 572 | |
a209dfc7 | 573 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
574 | if (!inode) |
575 | return NULL; | |
576 | ||
577 | sock = SOCKET_I(inode); | |
578 | ||
85fe4025 | 579 | inode->i_ino = get_next_ino(); |
89bddce5 | 580 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
581 | inode->i_uid = current_fsuid(); |
582 | inode->i_gid = current_fsgid(); | |
600e1779 | 583 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 584 | |
1da177e4 LT |
585 | return sock; |
586 | } | |
f4a00aac | 587 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 588 | |
1da177e4 | 589 | /** |
8a3c245c | 590 | * sock_release - close a socket |
1da177e4 LT |
591 | * @sock: socket to close |
592 | * | |
593 | * The socket is released from the protocol stack if it has a release | |
594 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 595 | * an inode not a file. |
1da177e4 | 596 | */ |
89bddce5 | 597 | |
6d8c50dc | 598 | static void __sock_release(struct socket *sock, struct inode *inode) |
1da177e4 LT |
599 | { |
600 | if (sock->ops) { | |
601 | struct module *owner = sock->ops->owner; | |
602 | ||
6d8c50dc CW |
603 | if (inode) |
604 | inode_lock(inode); | |
1da177e4 | 605 | sock->ops->release(sock); |
ff7b11aa | 606 | sock->sk = NULL; |
6d8c50dc CW |
607 | if (inode) |
608 | inode_unlock(inode); | |
1da177e4 LT |
609 | sock->ops = NULL; |
610 | module_put(owner); | |
611 | } | |
612 | ||
333f7909 | 613 | if (sock->wq.fasync_list) |
3410f22e | 614 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 615 | |
1da177e4 LT |
616 | if (!sock->file) { |
617 | iput(SOCK_INODE(sock)); | |
618 | return; | |
619 | } | |
89bddce5 | 620 | sock->file = NULL; |
1da177e4 | 621 | } |
6d8c50dc CW |
622 | |
623 | void sock_release(struct socket *sock) | |
624 | { | |
625 | __sock_release(sock, NULL); | |
626 | } | |
c6d409cf | 627 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 628 | |
c14ac945 | 629 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 630 | { |
140c55d4 ED |
631 | u8 flags = *tx_flags; |
632 | ||
c14ac945 | 633 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
634 | flags |= SKBTX_HW_TSTAMP; |
635 | ||
c14ac945 | 636 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
637 | flags |= SKBTX_SW_TSTAMP; |
638 | ||
c14ac945 | 639 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
640 | flags |= SKBTX_SCHED_TSTAMP; |
641 | ||
140c55d4 | 642 | *tx_flags = flags; |
20d49473 | 643 | } |
67cc0d40 | 644 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 645 | |
8c3c447b PA |
646 | INDIRECT_CALLABLE_DECLARE(int inet_sendmsg(struct socket *, struct msghdr *, |
647 | size_t)); | |
a648a592 PA |
648 | INDIRECT_CALLABLE_DECLARE(int inet6_sendmsg(struct socket *, struct msghdr *, |
649 | size_t)); | |
d8725c86 | 650 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 651 | { |
a648a592 PA |
652 | int ret = INDIRECT_CALL_INET(sock->ops->sendmsg, inet6_sendmsg, |
653 | inet_sendmsg, sock, msg, | |
654 | msg_data_left(msg)); | |
d8725c86 AV |
655 | BUG_ON(ret == -EIOCBQUEUED); |
656 | return ret; | |
1da177e4 LT |
657 | } |
658 | ||
85806af0 RD |
659 | /** |
660 | * sock_sendmsg - send a message through @sock | |
661 | * @sock: socket | |
662 | * @msg: message to send | |
663 | * | |
664 | * Sends @msg through @sock, passing through LSM. | |
665 | * Returns the number of bytes sent, or an error code. | |
666 | */ | |
d8725c86 | 667 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 668 | { |
d8725c86 | 669 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 670 | msg_data_left(msg)); |
228e548e | 671 | |
d8725c86 | 672 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 673 | } |
c6d409cf | 674 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 | 675 | |
8a3c245c PT |
676 | /** |
677 | * kernel_sendmsg - send a message through @sock (kernel-space) | |
678 | * @sock: socket | |
679 | * @msg: message header | |
680 | * @vec: kernel vec | |
681 | * @num: vec array length | |
682 | * @size: total message data size | |
683 | * | |
684 | * Builds the message data with @vec and sends it through @sock. | |
685 | * Returns the number of bytes sent, or an error code. | |
686 | */ | |
687 | ||
1da177e4 LT |
688 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
689 | struct kvec *vec, size_t num, size_t size) | |
690 | { | |
aa563d7b | 691 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
d8725c86 | 692 | return sock_sendmsg(sock, msg); |
1da177e4 | 693 | } |
c6d409cf | 694 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 695 | |
8a3c245c PT |
696 | /** |
697 | * kernel_sendmsg_locked - send a message through @sock (kernel-space) | |
698 | * @sk: sock | |
699 | * @msg: message header | |
700 | * @vec: output s/g array | |
701 | * @num: output s/g array length | |
702 | * @size: total message data size | |
703 | * | |
704 | * Builds the message data with @vec and sends it through @sock. | |
705 | * Returns the number of bytes sent, or an error code. | |
706 | * Caller must hold @sk. | |
707 | */ | |
708 | ||
306b13eb TH |
709 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
710 | struct kvec *vec, size_t num, size_t size) | |
711 | { | |
712 | struct socket *sock = sk->sk_socket; | |
713 | ||
714 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 715 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb | 716 | |
aa563d7b | 717 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
306b13eb TH |
718 | |
719 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
720 | } | |
721 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
722 | ||
8605330a SHY |
723 | static bool skb_is_err_queue(const struct sk_buff *skb) |
724 | { | |
725 | /* pkt_type of skbs enqueued on the error queue are set to | |
726 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
727 | * in recvmsg, since skbs received on a local socket will never | |
728 | * have a pkt_type of PACKET_OUTGOING. | |
729 | */ | |
730 | return skb->pkt_type == PACKET_OUTGOING; | |
731 | } | |
732 | ||
b50a5c70 ML |
733 | /* On transmit, software and hardware timestamps are returned independently. |
734 | * As the two skb clones share the hardware timestamp, which may be updated | |
735 | * before the software timestamp is received, a hardware TX timestamp may be | |
736 | * returned only if there is no software TX timestamp. Ignore false software | |
737 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
7f1bc6e9 | 738 | * option SO_TIMESTAMP_OLD(NS) is enabled on the socket, even when the skb has a |
b50a5c70 ML |
739 | * hardware timestamp. |
740 | */ | |
741 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
742 | { | |
743 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
744 | } | |
745 | ||
aad9c8c4 ML |
746 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb) |
747 | { | |
748 | struct scm_ts_pktinfo ts_pktinfo; | |
749 | struct net_device *orig_dev; | |
750 | ||
751 | if (!skb_mac_header_was_set(skb)) | |
752 | return; | |
753 | ||
754 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
755 | ||
756 | rcu_read_lock(); | |
757 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
758 | if (orig_dev) | |
759 | ts_pktinfo.if_index = orig_dev->ifindex; | |
760 | rcu_read_unlock(); | |
761 | ||
762 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
763 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
764 | sizeof(ts_pktinfo), &ts_pktinfo); | |
765 | } | |
766 | ||
92f37fd2 ED |
767 | /* |
768 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
769 | */ | |
770 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
771 | struct sk_buff *skb) | |
772 | { | |
20d49473 | 773 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
887feae3 | 774 | int new_tstamp = sock_flag(sk, SOCK_TSTAMP_NEW); |
9718475e DD |
775 | struct scm_timestamping_internal tss; |
776 | ||
b50a5c70 | 777 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
778 | struct skb_shared_hwtstamps *shhwtstamps = |
779 | skb_hwtstamps(skb); | |
780 | ||
781 | /* Race occurred between timestamp enabling and packet | |
782 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 783 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 784 | __net_timestamp(skb); |
b50a5c70 ML |
785 | false_tstamp = 1; |
786 | } | |
20d49473 PO |
787 | |
788 | if (need_software_tstamp) { | |
789 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
887feae3 DD |
790 | if (new_tstamp) { |
791 | struct __kernel_sock_timeval tv; | |
792 | ||
793 | skb_get_new_timestamp(skb, &tv); | |
794 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_NEW, | |
795 | sizeof(tv), &tv); | |
796 | } else { | |
797 | struct __kernel_old_timeval tv; | |
798 | ||
799 | skb_get_timestamp(skb, &tv); | |
800 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_OLD, | |
801 | sizeof(tv), &tv); | |
802 | } | |
20d49473 | 803 | } else { |
887feae3 DD |
804 | if (new_tstamp) { |
805 | struct __kernel_timespec ts; | |
806 | ||
807 | skb_get_new_timestampns(skb, &ts); | |
808 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_NEW, | |
809 | sizeof(ts), &ts); | |
810 | } else { | |
df1b4ba9 | 811 | struct __kernel_old_timespec ts; |
887feae3 DD |
812 | |
813 | skb_get_timestampns(skb, &ts); | |
814 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_OLD, | |
815 | sizeof(ts), &ts); | |
816 | } | |
20d49473 PO |
817 | } |
818 | } | |
819 | ||
f24b9be5 | 820 | memset(&tss, 0, sizeof(tss)); |
c199105d | 821 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
9718475e | 822 | ktime_to_timespec64_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 823 | empty = 0; |
4d276eb6 | 824 | if (shhwtstamps && |
b9f40e21 | 825 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
b50a5c70 | 826 | !skb_is_swtx_tstamp(skb, false_tstamp) && |
9718475e | 827 | ktime_to_timespec64_cond(shhwtstamps->hwtstamp, tss.ts + 2)) { |
4d276eb6 | 828 | empty = 0; |
aad9c8c4 ML |
829 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && |
830 | !skb_is_err_queue(skb)) | |
831 | put_ts_pktinfo(msg, skb); | |
832 | } | |
1c885808 | 833 | if (!empty) { |
9718475e DD |
834 | if (sock_flag(sk, SOCK_TSTAMP_NEW)) |
835 | put_cmsg_scm_timestamping64(msg, &tss); | |
836 | else | |
837 | put_cmsg_scm_timestamping(msg, &tss); | |
1c885808 | 838 | |
8605330a | 839 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 840 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
841 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
842 | skb->len, skb->data); | |
843 | } | |
92f37fd2 | 844 | } |
7c81fd8b ACM |
845 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
846 | ||
6e3e939f JB |
847 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
848 | struct sk_buff *skb) | |
849 | { | |
850 | int ack; | |
851 | ||
852 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
853 | return; | |
854 | if (!skb->wifi_acked_valid) | |
855 | return; | |
856 | ||
857 | ack = skb->wifi_acked; | |
858 | ||
859 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
860 | } | |
861 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
862 | ||
11165f14 | 863 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
864 | struct sk_buff *skb) | |
3b885787 | 865 | { |
744d5a3e | 866 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 867 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 868 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
869 | } |
870 | ||
767dd033 | 871 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
872 | struct sk_buff *skb) |
873 | { | |
874 | sock_recv_timestamp(msg, sk, skb); | |
875 | sock_recv_drops(msg, sk, skb); | |
876 | } | |
767dd033 | 877 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 878 | |
8c3c447b | 879 | INDIRECT_CALLABLE_DECLARE(int inet_recvmsg(struct socket *, struct msghdr *, |
a648a592 PA |
880 | size_t, int)); |
881 | INDIRECT_CALLABLE_DECLARE(int inet6_recvmsg(struct socket *, struct msghdr *, | |
882 | size_t, int)); | |
1b784140 | 883 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 884 | int flags) |
1da177e4 | 885 | { |
a648a592 PA |
886 | return INDIRECT_CALL_INET(sock->ops->recvmsg, inet6_recvmsg, |
887 | inet_recvmsg, sock, msg, msg_data_left(msg), | |
888 | flags); | |
1da177e4 LT |
889 | } |
890 | ||
85806af0 RD |
891 | /** |
892 | * sock_recvmsg - receive a message from @sock | |
893 | * @sock: socket | |
894 | * @msg: message to receive | |
895 | * @flags: message flags | |
896 | * | |
897 | * Receives @msg from @sock, passing through LSM. Returns the total number | |
898 | * of bytes received, or an error. | |
899 | */ | |
2da62906 | 900 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 901 | { |
2da62906 | 902 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 903 | |
2da62906 | 904 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 905 | } |
c6d409cf | 906 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 907 | |
c1249c0a | 908 | /** |
8a3c245c PT |
909 | * kernel_recvmsg - Receive a message from a socket (kernel space) |
910 | * @sock: The socket to receive the message from | |
911 | * @msg: Received message | |
912 | * @vec: Input s/g array for message data | |
913 | * @num: Size of input s/g array | |
914 | * @size: Number of bytes to read | |
915 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
c1249c0a | 916 | * |
8a3c245c PT |
917 | * On return the msg structure contains the scatter/gather array passed in the |
918 | * vec argument. The array is modified so that it consists of the unfilled | |
919 | * portion of the original array. | |
c1249c0a | 920 | * |
8a3c245c | 921 | * The returned value is the total number of bytes received, or an error. |
c1249c0a | 922 | */ |
8a3c245c | 923 | |
89bddce5 SH |
924 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
925 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
926 | { |
927 | mm_segment_t oldfs = get_fs(); | |
928 | int result; | |
929 | ||
aa563d7b | 930 | iov_iter_kvec(&msg->msg_iter, READ, vec, num, size); |
1da177e4 | 931 | set_fs(KERNEL_DS); |
2da62906 | 932 | result = sock_recvmsg(sock, msg, flags); |
1da177e4 LT |
933 | set_fs(oldfs); |
934 | return result; | |
935 | } | |
c6d409cf | 936 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 937 | |
ce1d4d3e CH |
938 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
939 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 940 | { |
1da177e4 LT |
941 | struct socket *sock; |
942 | int flags; | |
943 | ||
ce1d4d3e CH |
944 | sock = file->private_data; |
945 | ||
35f9c09f ED |
946 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
947 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
948 | flags |= more; | |
ce1d4d3e | 949 | |
e6949583 | 950 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 951 | } |
1da177e4 | 952 | |
9c55e01c | 953 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 954 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
955 | unsigned int flags) |
956 | { | |
957 | struct socket *sock = file->private_data; | |
958 | ||
997b37da | 959 | if (unlikely(!sock->ops->splice_read)) |
95506588 | 960 | return generic_file_splice_read(file, ppos, pipe, len, flags); |
997b37da | 961 | |
9c55e01c JA |
962 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
963 | } | |
964 | ||
8ae5e030 | 965 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 966 | { |
6d652330 AV |
967 | struct file *file = iocb->ki_filp; |
968 | struct socket *sock = file->private_data; | |
0345f931 | 969 | struct msghdr msg = {.msg_iter = *to, |
970 | .msg_iocb = iocb}; | |
8ae5e030 | 971 | ssize_t res; |
ce1d4d3e | 972 | |
ebfcd895 | 973 | if (file->f_flags & O_NONBLOCK || (iocb->ki_flags & IOCB_NOWAIT)) |
8ae5e030 AV |
974 | msg.msg_flags = MSG_DONTWAIT; |
975 | ||
976 | if (iocb->ki_pos != 0) | |
1da177e4 | 977 | return -ESPIPE; |
027445c3 | 978 | |
66ee59af | 979 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
980 | return 0; |
981 | ||
2da62906 | 982 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
983 | *to = msg.msg_iter; |
984 | return res; | |
1da177e4 LT |
985 | } |
986 | ||
8ae5e030 | 987 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 988 | { |
6d652330 AV |
989 | struct file *file = iocb->ki_filp; |
990 | struct socket *sock = file->private_data; | |
0345f931 | 991 | struct msghdr msg = {.msg_iter = *from, |
992 | .msg_iocb = iocb}; | |
8ae5e030 | 993 | ssize_t res; |
1da177e4 | 994 | |
8ae5e030 | 995 | if (iocb->ki_pos != 0) |
ce1d4d3e | 996 | return -ESPIPE; |
027445c3 | 997 | |
ebfcd895 | 998 | if (file->f_flags & O_NONBLOCK || (iocb->ki_flags & IOCB_NOWAIT)) |
8ae5e030 AV |
999 | msg.msg_flags = MSG_DONTWAIT; |
1000 | ||
6d652330 AV |
1001 | if (sock->type == SOCK_SEQPACKET) |
1002 | msg.msg_flags |= MSG_EOR; | |
1003 | ||
d8725c86 | 1004 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
1005 | *from = msg.msg_iter; |
1006 | return res; | |
1da177e4 LT |
1007 | } |
1008 | ||
1da177e4 LT |
1009 | /* |
1010 | * Atomic setting of ioctl hooks to avoid race | |
1011 | * with module unload. | |
1012 | */ | |
1013 | ||
4a3e2f71 | 1014 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 1015 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 1016 | |
881d966b | 1017 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 1018 | { |
4a3e2f71 | 1019 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 1020 | br_ioctl_hook = hook; |
4a3e2f71 | 1021 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1022 | } |
1023 | EXPORT_SYMBOL(brioctl_set); | |
1024 | ||
4a3e2f71 | 1025 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 1026 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 1027 | |
881d966b | 1028 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 1029 | { |
4a3e2f71 | 1030 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1031 | vlan_ioctl_hook = hook; |
4a3e2f71 | 1032 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
1033 | } |
1034 | EXPORT_SYMBOL(vlan_ioctl_set); | |
1035 | ||
4a3e2f71 | 1036 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 1037 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 1038 | |
89bddce5 | 1039 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 1040 | { |
4a3e2f71 | 1041 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 1042 | dlci_ioctl_hook = hook; |
4a3e2f71 | 1043 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
1044 | } |
1045 | EXPORT_SYMBOL(dlci_ioctl_set); | |
1046 | ||
6b96018b | 1047 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
63ff03ab | 1048 | unsigned int cmd, unsigned long arg) |
6b96018b AB |
1049 | { |
1050 | int err; | |
1051 | void __user *argp = (void __user *)arg; | |
1052 | ||
1053 | err = sock->ops->ioctl(sock, cmd, arg); | |
1054 | ||
1055 | /* | |
1056 | * If this ioctl is unknown try to hand it down | |
1057 | * to the NIC driver. | |
1058 | */ | |
36fd633e AV |
1059 | if (err != -ENOIOCTLCMD) |
1060 | return err; | |
6b96018b | 1061 | |
36fd633e AV |
1062 | if (cmd == SIOCGIFCONF) { |
1063 | struct ifconf ifc; | |
1064 | if (copy_from_user(&ifc, argp, sizeof(struct ifconf))) | |
1065 | return -EFAULT; | |
1066 | rtnl_lock(); | |
1067 | err = dev_ifconf(net, &ifc, sizeof(struct ifreq)); | |
1068 | rtnl_unlock(); | |
1069 | if (!err && copy_to_user(argp, &ifc, sizeof(struct ifconf))) | |
1070 | err = -EFAULT; | |
44c02a2c AV |
1071 | } else { |
1072 | struct ifreq ifr; | |
1073 | bool need_copyout; | |
63ff03ab | 1074 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) |
44c02a2c AV |
1075 | return -EFAULT; |
1076 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1077 | if (!err && need_copyout) | |
63ff03ab | 1078 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) |
44c02a2c | 1079 | return -EFAULT; |
36fd633e | 1080 | } |
6b96018b AB |
1081 | return err; |
1082 | } | |
1083 | ||
1da177e4 LT |
1084 | /* |
1085 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1086 | * what to do with it - that's up to the protocol still. | |
1087 | */ | |
1088 | ||
8a3c245c PT |
1089 | /** |
1090 | * get_net_ns - increment the refcount of the network namespace | |
1091 | * @ns: common namespace (net) | |
1092 | * | |
1093 | * Returns the net's common namespace. | |
1094 | */ | |
1095 | ||
d8d211a2 | 1096 | struct ns_common *get_net_ns(struct ns_common *ns) |
c62cce2c AV |
1097 | { |
1098 | return &get_net(container_of(ns, struct net, ns))->ns; | |
1099 | } | |
d8d211a2 | 1100 | EXPORT_SYMBOL_GPL(get_net_ns); |
c62cce2c | 1101 | |
1da177e4 LT |
1102 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
1103 | { | |
1104 | struct socket *sock; | |
881d966b | 1105 | struct sock *sk; |
1da177e4 LT |
1106 | void __user *argp = (void __user *)arg; |
1107 | int pid, err; | |
881d966b | 1108 | struct net *net; |
1da177e4 | 1109 | |
b69aee04 | 1110 | sock = file->private_data; |
881d966b | 1111 | sk = sock->sk; |
3b1e0a65 | 1112 | net = sock_net(sk); |
44c02a2c AV |
1113 | if (unlikely(cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))) { |
1114 | struct ifreq ifr; | |
1115 | bool need_copyout; | |
1116 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
1117 | return -EFAULT; | |
1118 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1119 | if (!err && need_copyout) | |
1120 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
1121 | return -EFAULT; | |
1da177e4 | 1122 | } else |
3d23e349 | 1123 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1124 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
b1b0c245 | 1125 | err = wext_handle_ioctl(net, cmd, argp); |
1da177e4 | 1126 | } else |
3d23e349 | 1127 | #endif |
89bddce5 | 1128 | switch (cmd) { |
1da177e4 LT |
1129 | case FIOSETOWN: |
1130 | case SIOCSPGRP: | |
1131 | err = -EFAULT; | |
1132 | if (get_user(pid, (int __user *)argp)) | |
1133 | break; | |
393cc3f5 | 1134 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1135 | break; |
1136 | case FIOGETOWN: | |
1137 | case SIOCGPGRP: | |
609d7fa9 | 1138 | err = put_user(f_getown(sock->file), |
89bddce5 | 1139 | (int __user *)argp); |
1da177e4 LT |
1140 | break; |
1141 | case SIOCGIFBR: | |
1142 | case SIOCSIFBR: | |
1143 | case SIOCBRADDBR: | |
1144 | case SIOCBRDELBR: | |
1145 | err = -ENOPKG; | |
1146 | if (!br_ioctl_hook) | |
1147 | request_module("bridge"); | |
1148 | ||
4a3e2f71 | 1149 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1150 | if (br_ioctl_hook) |
881d966b | 1151 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1152 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1153 | break; |
1154 | case SIOCGIFVLAN: | |
1155 | case SIOCSIFVLAN: | |
1156 | err = -ENOPKG; | |
1157 | if (!vlan_ioctl_hook) | |
1158 | request_module("8021q"); | |
1159 | ||
4a3e2f71 | 1160 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1161 | if (vlan_ioctl_hook) |
881d966b | 1162 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1163 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1164 | break; |
1da177e4 LT |
1165 | case SIOCADDDLCI: |
1166 | case SIOCDELDLCI: | |
1167 | err = -ENOPKG; | |
1168 | if (!dlci_ioctl_hook) | |
1169 | request_module("dlci"); | |
1170 | ||
7512cbf6 PE |
1171 | mutex_lock(&dlci_ioctl_mutex); |
1172 | if (dlci_ioctl_hook) | |
1da177e4 | 1173 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1174 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 | 1175 | break; |
c62cce2c AV |
1176 | case SIOCGSKNS: |
1177 | err = -EPERM; | |
1178 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1179 | break; | |
1180 | ||
1181 | err = open_related_ns(&net->ns, get_net_ns); | |
1182 | break; | |
0768e170 AB |
1183 | case SIOCGSTAMP_OLD: |
1184 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
1185 | if (!sock->ops->gettstamp) { |
1186 | err = -ENOIOCTLCMD; | |
1187 | break; | |
1188 | } | |
1189 | err = sock->ops->gettstamp(sock, argp, | |
0768e170 AB |
1190 | cmd == SIOCGSTAMP_OLD, |
1191 | !IS_ENABLED(CONFIG_64BIT)); | |
60747828 | 1192 | break; |
0768e170 AB |
1193 | case SIOCGSTAMP_NEW: |
1194 | case SIOCGSTAMPNS_NEW: | |
1195 | if (!sock->ops->gettstamp) { | |
1196 | err = -ENOIOCTLCMD; | |
1197 | break; | |
1198 | } | |
1199 | err = sock->ops->gettstamp(sock, argp, | |
1200 | cmd == SIOCGSTAMP_NEW, | |
1201 | false); | |
c7cbdbf2 | 1202 | break; |
1da177e4 | 1203 | default: |
63ff03ab | 1204 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1205 | break; |
89bddce5 | 1206 | } |
1da177e4 LT |
1207 | return err; |
1208 | } | |
1209 | ||
8a3c245c PT |
1210 | /** |
1211 | * sock_create_lite - creates a socket | |
1212 | * @family: protocol family (AF_INET, ...) | |
1213 | * @type: communication type (SOCK_STREAM, ...) | |
1214 | * @protocol: protocol (0, ...) | |
1215 | * @res: new socket | |
1216 | * | |
1217 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1218 | * The new socket initialization is not complete, see kernel_accept(). | |
1219 | * Returns 0 or an error. On failure @res is set to %NULL. | |
1220 | * This function internally uses GFP_KERNEL. | |
1221 | */ | |
1222 | ||
1da177e4 LT |
1223 | int sock_create_lite(int family, int type, int protocol, struct socket **res) |
1224 | { | |
1225 | int err; | |
1226 | struct socket *sock = NULL; | |
89bddce5 | 1227 | |
1da177e4 LT |
1228 | err = security_socket_create(family, type, protocol, 1); |
1229 | if (err) | |
1230 | goto out; | |
1231 | ||
1232 | sock = sock_alloc(); | |
1233 | if (!sock) { | |
1234 | err = -ENOMEM; | |
1235 | goto out; | |
1236 | } | |
1237 | ||
1da177e4 | 1238 | sock->type = type; |
7420ed23 VY |
1239 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1240 | if (err) | |
1241 | goto out_release; | |
1242 | ||
1da177e4 LT |
1243 | out: |
1244 | *res = sock; | |
1245 | return err; | |
7420ed23 VY |
1246 | out_release: |
1247 | sock_release(sock); | |
1248 | sock = NULL; | |
1249 | goto out; | |
1da177e4 | 1250 | } |
c6d409cf | 1251 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1252 | |
1253 | /* No kernel lock held - perfect */ | |
ade994f4 | 1254 | static __poll_t sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1255 | { |
3cafb376 | 1256 | struct socket *sock = file->private_data; |
a331de3b | 1257 | __poll_t events = poll_requested_events(wait), flag = 0; |
2d48d67f | 1258 | |
e88958e6 CH |
1259 | if (!sock->ops->poll) |
1260 | return 0; | |
f641f13b | 1261 | |
a331de3b CH |
1262 | if (sk_can_busy_loop(sock->sk)) { |
1263 | /* poll once if requested by the syscall */ | |
1264 | if (events & POLL_BUSY_LOOP) | |
1265 | sk_busy_loop(sock->sk, 1); | |
1266 | ||
1267 | /* if this socket can poll_ll, tell the system call */ | |
1268 | flag = POLL_BUSY_LOOP; | |
1269 | } | |
1270 | ||
1271 | return sock->ops->poll(file, sock, wait) | flag; | |
1da177e4 LT |
1272 | } |
1273 | ||
89bddce5 | 1274 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1275 | { |
b69aee04 | 1276 | struct socket *sock = file->private_data; |
1da177e4 LT |
1277 | |
1278 | return sock->ops->mmap(file, sock, vma); | |
1279 | } | |
1280 | ||
20380731 | 1281 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1282 | { |
6d8c50dc | 1283 | __sock_release(SOCKET_I(inode), inode); |
1da177e4 LT |
1284 | return 0; |
1285 | } | |
1286 | ||
1287 | /* | |
1288 | * Update the socket async list | |
1289 | * | |
1290 | * Fasync_list locking strategy. | |
1291 | * | |
1292 | * 1. fasync_list is modified only under process context socket lock | |
1293 | * i.e. under semaphore. | |
1294 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1295 | * or under socket lock |
1da177e4 LT |
1296 | */ |
1297 | ||
1298 | static int sock_fasync(int fd, struct file *filp, int on) | |
1299 | { | |
989a2979 ED |
1300 | struct socket *sock = filp->private_data; |
1301 | struct sock *sk = sock->sk; | |
333f7909 | 1302 | struct socket_wq *wq = &sock->wq; |
1da177e4 | 1303 | |
989a2979 | 1304 | if (sk == NULL) |
1da177e4 | 1305 | return -EINVAL; |
1da177e4 LT |
1306 | |
1307 | lock_sock(sk); | |
eaefd110 | 1308 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1309 | |
eaefd110 | 1310 | if (!wq->fasync_list) |
989a2979 ED |
1311 | sock_reset_flag(sk, SOCK_FASYNC); |
1312 | else | |
bcdce719 | 1313 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1314 | |
989a2979 | 1315 | release_sock(sk); |
1da177e4 LT |
1316 | return 0; |
1317 | } | |
1318 | ||
ceb5d58b | 1319 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1320 | |
ceb5d58b | 1321 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1322 | { |
ceb5d58b | 1323 | if (!wq || !wq->fasync_list) |
1da177e4 | 1324 | return -1; |
ceb5d58b | 1325 | |
89bddce5 | 1326 | switch (how) { |
8d8ad9d7 | 1327 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1328 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1329 | break; |
1330 | goto call_kill; | |
8d8ad9d7 | 1331 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1332 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 LT |
1333 | break; |
1334 | /* fall through */ | |
8d8ad9d7 | 1335 | case SOCK_WAKE_IO: |
89bddce5 | 1336 | call_kill: |
43815482 | 1337 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1338 | break; |
8d8ad9d7 | 1339 | case SOCK_WAKE_URG: |
43815482 | 1340 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1341 | } |
ceb5d58b | 1342 | |
1da177e4 LT |
1343 | return 0; |
1344 | } | |
c6d409cf | 1345 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1346 | |
8a3c245c PT |
1347 | /** |
1348 | * __sock_create - creates a socket | |
1349 | * @net: net namespace | |
1350 | * @family: protocol family (AF_INET, ...) | |
1351 | * @type: communication type (SOCK_STREAM, ...) | |
1352 | * @protocol: protocol (0, ...) | |
1353 | * @res: new socket | |
1354 | * @kern: boolean for kernel space sockets | |
1355 | * | |
1356 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1357 | * Returns 0 or an error. On failure @res is set to %NULL. @kern must | |
1358 | * be set to true if the socket resides in kernel space. | |
1359 | * This function internally uses GFP_KERNEL. | |
1360 | */ | |
1361 | ||
721db93a | 1362 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1363 | struct socket **res, int kern) |
1da177e4 LT |
1364 | { |
1365 | int err; | |
1366 | struct socket *sock; | |
55737fda | 1367 | const struct net_proto_family *pf; |
1da177e4 LT |
1368 | |
1369 | /* | |
89bddce5 | 1370 | * Check protocol is in range |
1da177e4 LT |
1371 | */ |
1372 | if (family < 0 || family >= NPROTO) | |
1373 | return -EAFNOSUPPORT; | |
1374 | if (type < 0 || type >= SOCK_MAX) | |
1375 | return -EINVAL; | |
1376 | ||
1377 | /* Compatibility. | |
1378 | ||
1379 | This uglymoron is moved from INET layer to here to avoid | |
1380 | deadlock in module load. | |
1381 | */ | |
1382 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1383 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1384 | current->comm); | |
1da177e4 LT |
1385 | family = PF_PACKET; |
1386 | } | |
1387 | ||
1388 | err = security_socket_create(family, type, protocol, kern); | |
1389 | if (err) | |
1390 | return err; | |
89bddce5 | 1391 | |
55737fda SH |
1392 | /* |
1393 | * Allocate the socket and allow the family to set things up. if | |
1394 | * the protocol is 0, the family is instructed to select an appropriate | |
1395 | * default. | |
1396 | */ | |
1397 | sock = sock_alloc(); | |
1398 | if (!sock) { | |
e87cc472 | 1399 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1400 | return -ENFILE; /* Not exactly a match, but its the |
1401 | closest posix thing */ | |
1402 | } | |
1403 | ||
1404 | sock->type = type; | |
1405 | ||
95a5afca | 1406 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1407 | /* Attempt to load a protocol module if the find failed. |
1408 | * | |
1409 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1410 | * requested real, full-featured networking support upon configuration. |
1411 | * Otherwise module support will break! | |
1412 | */ | |
190683a9 | 1413 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1414 | request_module("net-pf-%d", family); |
1da177e4 LT |
1415 | #endif |
1416 | ||
55737fda SH |
1417 | rcu_read_lock(); |
1418 | pf = rcu_dereference(net_families[family]); | |
1419 | err = -EAFNOSUPPORT; | |
1420 | if (!pf) | |
1421 | goto out_release; | |
1da177e4 LT |
1422 | |
1423 | /* | |
1424 | * We will call the ->create function, that possibly is in a loadable | |
1425 | * module, so we have to bump that loadable module refcnt first. | |
1426 | */ | |
55737fda | 1427 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1428 | goto out_release; |
1429 | ||
55737fda SH |
1430 | /* Now protected by module ref count */ |
1431 | rcu_read_unlock(); | |
1432 | ||
3f378b68 | 1433 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1434 | if (err < 0) |
1da177e4 | 1435 | goto out_module_put; |
a79af59e | 1436 | |
1da177e4 LT |
1437 | /* |
1438 | * Now to bump the refcnt of the [loadable] module that owns this | |
1439 | * socket at sock_release time we decrement its refcnt. | |
1440 | */ | |
55737fda SH |
1441 | if (!try_module_get(sock->ops->owner)) |
1442 | goto out_module_busy; | |
1443 | ||
1da177e4 LT |
1444 | /* |
1445 | * Now that we're done with the ->create function, the [loadable] | |
1446 | * module can have its refcnt decremented | |
1447 | */ | |
55737fda | 1448 | module_put(pf->owner); |
7420ed23 VY |
1449 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1450 | if (err) | |
3b185525 | 1451 | goto out_sock_release; |
55737fda | 1452 | *res = sock; |
1da177e4 | 1453 | |
55737fda SH |
1454 | return 0; |
1455 | ||
1456 | out_module_busy: | |
1457 | err = -EAFNOSUPPORT; | |
1da177e4 | 1458 | out_module_put: |
55737fda SH |
1459 | sock->ops = NULL; |
1460 | module_put(pf->owner); | |
1461 | out_sock_release: | |
1da177e4 | 1462 | sock_release(sock); |
55737fda SH |
1463 | return err; |
1464 | ||
1465 | out_release: | |
1466 | rcu_read_unlock(); | |
1467 | goto out_sock_release; | |
1da177e4 | 1468 | } |
721db93a | 1469 | EXPORT_SYMBOL(__sock_create); |
1da177e4 | 1470 | |
8a3c245c PT |
1471 | /** |
1472 | * sock_create - creates a socket | |
1473 | * @family: protocol family (AF_INET, ...) | |
1474 | * @type: communication type (SOCK_STREAM, ...) | |
1475 | * @protocol: protocol (0, ...) | |
1476 | * @res: new socket | |
1477 | * | |
1478 | * A wrapper around __sock_create(). | |
1479 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1480 | */ | |
1481 | ||
1da177e4 LT |
1482 | int sock_create(int family, int type, int protocol, struct socket **res) |
1483 | { | |
1b8d7ae4 | 1484 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1485 | } |
c6d409cf | 1486 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1487 | |
8a3c245c PT |
1488 | /** |
1489 | * sock_create_kern - creates a socket (kernel space) | |
1490 | * @net: net namespace | |
1491 | * @family: protocol family (AF_INET, ...) | |
1492 | * @type: communication type (SOCK_STREAM, ...) | |
1493 | * @protocol: protocol (0, ...) | |
1494 | * @res: new socket | |
1495 | * | |
1496 | * A wrapper around __sock_create(). | |
1497 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1498 | */ | |
1499 | ||
eeb1bd5c | 1500 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1501 | { |
eeb1bd5c | 1502 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1503 | } |
c6d409cf | 1504 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1505 | |
9d6a15c3 | 1506 | int __sys_socket(int family, int type, int protocol) |
1da177e4 LT |
1507 | { |
1508 | int retval; | |
1509 | struct socket *sock; | |
a677a039 UD |
1510 | int flags; |
1511 | ||
e38b36f3 UD |
1512 | /* Check the SOCK_* constants for consistency. */ |
1513 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1514 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1515 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1516 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1517 | ||
a677a039 | 1518 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1519 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1520 | return -EINVAL; |
1521 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1522 | |
aaca0bdc UD |
1523 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1524 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1525 | ||
1da177e4 LT |
1526 | retval = sock_create(family, type, protocol, &sock); |
1527 | if (retval < 0) | |
8e1611e2 | 1528 | return retval; |
1da177e4 | 1529 | |
8e1611e2 | 1530 | return sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1531 | } |
1532 | ||
9d6a15c3 DB |
1533 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1534 | { | |
1535 | return __sys_socket(family, type, protocol); | |
1536 | } | |
1537 | ||
1da177e4 LT |
1538 | /* |
1539 | * Create a pair of connected sockets. | |
1540 | */ | |
1541 | ||
6debc8d8 | 1542 | int __sys_socketpair(int family, int type, int protocol, int __user *usockvec) |
1da177e4 LT |
1543 | { |
1544 | struct socket *sock1, *sock2; | |
1545 | int fd1, fd2, err; | |
db349509 | 1546 | struct file *newfile1, *newfile2; |
a677a039 UD |
1547 | int flags; |
1548 | ||
1549 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1550 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1551 | return -EINVAL; |
1552 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1553 | |
aaca0bdc UD |
1554 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1555 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1556 | ||
016a266b AV |
1557 | /* |
1558 | * reserve descriptors and make sure we won't fail | |
1559 | * to return them to userland. | |
1560 | */ | |
1561 | fd1 = get_unused_fd_flags(flags); | |
1562 | if (unlikely(fd1 < 0)) | |
1563 | return fd1; | |
1564 | ||
1565 | fd2 = get_unused_fd_flags(flags); | |
1566 | if (unlikely(fd2 < 0)) { | |
1567 | put_unused_fd(fd1); | |
1568 | return fd2; | |
1569 | } | |
1570 | ||
1571 | err = put_user(fd1, &usockvec[0]); | |
1572 | if (err) | |
1573 | goto out; | |
1574 | ||
1575 | err = put_user(fd2, &usockvec[1]); | |
1576 | if (err) | |
1577 | goto out; | |
1578 | ||
1da177e4 LT |
1579 | /* |
1580 | * Obtain the first socket and check if the underlying protocol | |
1581 | * supports the socketpair call. | |
1582 | */ | |
1583 | ||
1584 | err = sock_create(family, type, protocol, &sock1); | |
016a266b | 1585 | if (unlikely(err < 0)) |
1da177e4 LT |
1586 | goto out; |
1587 | ||
1588 | err = sock_create(family, type, protocol, &sock2); | |
016a266b AV |
1589 | if (unlikely(err < 0)) { |
1590 | sock_release(sock1); | |
1591 | goto out; | |
bf3c23d1 | 1592 | } |
d73aa286 | 1593 | |
d47cd945 DH |
1594 | err = security_socket_socketpair(sock1, sock2); |
1595 | if (unlikely(err)) { | |
1596 | sock_release(sock2); | |
1597 | sock_release(sock1); | |
1598 | goto out; | |
1599 | } | |
1600 | ||
016a266b AV |
1601 | err = sock1->ops->socketpair(sock1, sock2); |
1602 | if (unlikely(err < 0)) { | |
1603 | sock_release(sock2); | |
1604 | sock_release(sock1); | |
1605 | goto out; | |
28407630 AV |
1606 | } |
1607 | ||
aab174f0 | 1608 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1609 | if (IS_ERR(newfile1)) { |
28407630 | 1610 | err = PTR_ERR(newfile1); |
016a266b AV |
1611 | sock_release(sock2); |
1612 | goto out; | |
28407630 AV |
1613 | } |
1614 | ||
aab174f0 | 1615 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1616 | if (IS_ERR(newfile2)) { |
1617 | err = PTR_ERR(newfile2); | |
016a266b AV |
1618 | fput(newfile1); |
1619 | goto out; | |
db349509 AV |
1620 | } |
1621 | ||
157cf649 | 1622 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1623 | |
db349509 AV |
1624 | fd_install(fd1, newfile1); |
1625 | fd_install(fd2, newfile2); | |
d73aa286 | 1626 | return 0; |
1da177e4 | 1627 | |
016a266b | 1628 | out: |
d73aa286 | 1629 | put_unused_fd(fd2); |
d73aa286 | 1630 | put_unused_fd(fd1); |
1da177e4 LT |
1631 | return err; |
1632 | } | |
1633 | ||
6debc8d8 DB |
1634 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1635 | int __user *, usockvec) | |
1636 | { | |
1637 | return __sys_socketpair(family, type, protocol, usockvec); | |
1638 | } | |
1639 | ||
1da177e4 LT |
1640 | /* |
1641 | * Bind a name to a socket. Nothing much to do here since it's | |
1642 | * the protocol's responsibility to handle the local address. | |
1643 | * | |
1644 | * We move the socket address to kernel space before we call | |
1645 | * the protocol layer (having also checked the address is ok). | |
1646 | */ | |
1647 | ||
a87d35d8 | 1648 | int __sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
1da177e4 LT |
1649 | { |
1650 | struct socket *sock; | |
230b1839 | 1651 | struct sockaddr_storage address; |
6cb153ca | 1652 | int err, fput_needed; |
1da177e4 | 1653 | |
89bddce5 | 1654 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1655 | if (sock) { |
43db362d | 1656 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
068b88cc | 1657 | if (!err) { |
89bddce5 | 1658 | err = security_socket_bind(sock, |
230b1839 | 1659 | (struct sockaddr *)&address, |
89bddce5 | 1660 | addrlen); |
6cb153ca BL |
1661 | if (!err) |
1662 | err = sock->ops->bind(sock, | |
89bddce5 | 1663 | (struct sockaddr *) |
230b1839 | 1664 | &address, addrlen); |
1da177e4 | 1665 | } |
6cb153ca | 1666 | fput_light(sock->file, fput_needed); |
89bddce5 | 1667 | } |
1da177e4 LT |
1668 | return err; |
1669 | } | |
1670 | ||
a87d35d8 DB |
1671 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1672 | { | |
1673 | return __sys_bind(fd, umyaddr, addrlen); | |
1674 | } | |
1675 | ||
1da177e4 LT |
1676 | /* |
1677 | * Perform a listen. Basically, we allow the protocol to do anything | |
1678 | * necessary for a listen, and if that works, we mark the socket as | |
1679 | * ready for listening. | |
1680 | */ | |
1681 | ||
25e290ee | 1682 | int __sys_listen(int fd, int backlog) |
1da177e4 LT |
1683 | { |
1684 | struct socket *sock; | |
6cb153ca | 1685 | int err, fput_needed; |
b8e1f9b5 | 1686 | int somaxconn; |
89bddce5 SH |
1687 | |
1688 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1689 | if (sock) { | |
8efa6e93 | 1690 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1691 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1692 | backlog = somaxconn; |
1da177e4 LT |
1693 | |
1694 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1695 | if (!err) |
1696 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1697 | |
6cb153ca | 1698 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1699 | } |
1700 | return err; | |
1701 | } | |
1702 | ||
25e290ee DB |
1703 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1704 | { | |
1705 | return __sys_listen(fd, backlog); | |
1706 | } | |
1707 | ||
de2ea4b6 JA |
1708 | int __sys_accept4_file(struct file *file, unsigned file_flags, |
1709 | struct sockaddr __user *upeer_sockaddr, | |
09952e3e JA |
1710 | int __user *upeer_addrlen, int flags, |
1711 | unsigned long nofile) | |
1da177e4 LT |
1712 | { |
1713 | struct socket *sock, *newsock; | |
39d8c1b6 | 1714 | struct file *newfile; |
de2ea4b6 | 1715 | int err, len, newfd; |
230b1839 | 1716 | struct sockaddr_storage address; |
1da177e4 | 1717 | |
77d27200 | 1718 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1719 | return -EINVAL; |
1720 | ||
1721 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1722 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1723 | ||
de2ea4b6 | 1724 | sock = sock_from_file(file, &err); |
1da177e4 LT |
1725 | if (!sock) |
1726 | goto out; | |
1727 | ||
1728 | err = -ENFILE; | |
c6d409cf ED |
1729 | newsock = sock_alloc(); |
1730 | if (!newsock) | |
de2ea4b6 | 1731 | goto out; |
1da177e4 LT |
1732 | |
1733 | newsock->type = sock->type; | |
1734 | newsock->ops = sock->ops; | |
1735 | ||
1da177e4 LT |
1736 | /* |
1737 | * We don't need try_module_get here, as the listening socket (sock) | |
1738 | * has the protocol module (sock->ops->owner) held. | |
1739 | */ | |
1740 | __module_get(newsock->ops->owner); | |
1741 | ||
09952e3e | 1742 | newfd = __get_unused_fd_flags(flags, nofile); |
39d8c1b6 DM |
1743 | if (unlikely(newfd < 0)) { |
1744 | err = newfd; | |
9a1875e6 | 1745 | sock_release(newsock); |
de2ea4b6 | 1746 | goto out; |
39d8c1b6 | 1747 | } |
aab174f0 | 1748 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1749 | if (IS_ERR(newfile)) { |
28407630 AV |
1750 | err = PTR_ERR(newfile); |
1751 | put_unused_fd(newfd); | |
de2ea4b6 | 1752 | goto out; |
28407630 | 1753 | } |
39d8c1b6 | 1754 | |
a79af59e FF |
1755 | err = security_socket_accept(sock, newsock); |
1756 | if (err) | |
39d8c1b6 | 1757 | goto out_fd; |
a79af59e | 1758 | |
de2ea4b6 JA |
1759 | err = sock->ops->accept(sock, newsock, sock->file->f_flags | file_flags, |
1760 | false); | |
1da177e4 | 1761 | if (err < 0) |
39d8c1b6 | 1762 | goto out_fd; |
1da177e4 LT |
1763 | |
1764 | if (upeer_sockaddr) { | |
9b2c45d4 DV |
1765 | len = newsock->ops->getname(newsock, |
1766 | (struct sockaddr *)&address, 2); | |
1767 | if (len < 0) { | |
1da177e4 | 1768 | err = -ECONNABORTED; |
39d8c1b6 | 1769 | goto out_fd; |
1da177e4 | 1770 | } |
43db362d | 1771 | err = move_addr_to_user(&address, |
230b1839 | 1772 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1773 | if (err < 0) |
39d8c1b6 | 1774 | goto out_fd; |
1da177e4 LT |
1775 | } |
1776 | ||
1777 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1778 | ||
39d8c1b6 DM |
1779 | fd_install(newfd, newfile); |
1780 | err = newfd; | |
1da177e4 LT |
1781 | out: |
1782 | return err; | |
39d8c1b6 | 1783 | out_fd: |
9606a216 | 1784 | fput(newfile); |
39d8c1b6 | 1785 | put_unused_fd(newfd); |
de2ea4b6 JA |
1786 | goto out; |
1787 | ||
1788 | } | |
1789 | ||
1790 | /* | |
1791 | * For accept, we attempt to create a new socket, set up the link | |
1792 | * with the client, wake up the client, then return the new | |
1793 | * connected fd. We collect the address of the connector in kernel | |
1794 | * space and move it to user at the very end. This is unclean because | |
1795 | * we open the socket then return an error. | |
1796 | * | |
1797 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1798 | * status to recvmsg. We need to add that support in a way thats | |
1799 | * clean when we restructure accept also. | |
1800 | */ | |
1801 | ||
1802 | int __sys_accept4(int fd, struct sockaddr __user *upeer_sockaddr, | |
1803 | int __user *upeer_addrlen, int flags) | |
1804 | { | |
1805 | int ret = -EBADF; | |
1806 | struct fd f; | |
1807 | ||
1808 | f = fdget(fd); | |
1809 | if (f.file) { | |
1810 | ret = __sys_accept4_file(f.file, 0, upeer_sockaddr, | |
09952e3e JA |
1811 | upeer_addrlen, flags, |
1812 | rlimit(RLIMIT_NOFILE)); | |
de2ea4b6 JA |
1813 | if (f.flags) |
1814 | fput(f.file); | |
1815 | } | |
1816 | ||
1817 | return ret; | |
1da177e4 LT |
1818 | } |
1819 | ||
4541e805 DB |
1820 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1821 | int __user *, upeer_addrlen, int, flags) | |
1822 | { | |
1823 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, flags); | |
1824 | } | |
1825 | ||
20f37034 HC |
1826 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1827 | int __user *, upeer_addrlen) | |
aaca0bdc | 1828 | { |
4541e805 | 1829 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1830 | } |
1831 | ||
1da177e4 LT |
1832 | /* |
1833 | * Attempt to connect to a socket with the server address. The address | |
1834 | * is in user space so we verify it is OK and move it to kernel space. | |
1835 | * | |
1836 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1837 | * break bindings | |
1838 | * | |
1839 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1840 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1841 | * include the -EINPROGRESS status for such sockets. | |
1842 | */ | |
1843 | ||
f499a021 | 1844 | int __sys_connect_file(struct file *file, struct sockaddr_storage *address, |
bd3ded31 | 1845 | int addrlen, int file_flags) |
1da177e4 LT |
1846 | { |
1847 | struct socket *sock; | |
bd3ded31 | 1848 | int err; |
1da177e4 | 1849 | |
bd3ded31 | 1850 | sock = sock_from_file(file, &err); |
1da177e4 LT |
1851 | if (!sock) |
1852 | goto out; | |
1da177e4 | 1853 | |
89bddce5 | 1854 | err = |
f499a021 | 1855 | security_socket_connect(sock, (struct sockaddr *)address, addrlen); |
1da177e4 | 1856 | if (err) |
bd3ded31 | 1857 | goto out; |
1da177e4 | 1858 | |
f499a021 | 1859 | err = sock->ops->connect(sock, (struct sockaddr *)address, addrlen, |
bd3ded31 | 1860 | sock->file->f_flags | file_flags); |
1da177e4 LT |
1861 | out: |
1862 | return err; | |
1863 | } | |
1864 | ||
bd3ded31 JA |
1865 | int __sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
1866 | { | |
1867 | int ret = -EBADF; | |
1868 | struct fd f; | |
1869 | ||
1870 | f = fdget(fd); | |
1871 | if (f.file) { | |
f499a021 JA |
1872 | struct sockaddr_storage address; |
1873 | ||
1874 | ret = move_addr_to_kernel(uservaddr, addrlen, &address); | |
1875 | if (!ret) | |
1876 | ret = __sys_connect_file(f.file, &address, addrlen, 0); | |
bd3ded31 JA |
1877 | if (f.flags) |
1878 | fput(f.file); | |
1879 | } | |
1880 | ||
1881 | return ret; | |
1882 | } | |
1883 | ||
1387c2c2 DB |
1884 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1885 | int, addrlen) | |
1886 | { | |
1887 | return __sys_connect(fd, uservaddr, addrlen); | |
1888 | } | |
1889 | ||
1da177e4 LT |
1890 | /* |
1891 | * Get the local address ('name') of a socket object. Move the obtained | |
1892 | * name to user space. | |
1893 | */ | |
1894 | ||
8882a107 DB |
1895 | int __sys_getsockname(int fd, struct sockaddr __user *usockaddr, |
1896 | int __user *usockaddr_len) | |
1da177e4 LT |
1897 | { |
1898 | struct socket *sock; | |
230b1839 | 1899 | struct sockaddr_storage address; |
9b2c45d4 | 1900 | int err, fput_needed; |
89bddce5 | 1901 | |
6cb153ca | 1902 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1903 | if (!sock) |
1904 | goto out; | |
1905 | ||
1906 | err = security_socket_getsockname(sock); | |
1907 | if (err) | |
1908 | goto out_put; | |
1909 | ||
9b2c45d4 DV |
1910 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 0); |
1911 | if (err < 0) | |
1da177e4 | 1912 | goto out_put; |
9b2c45d4 DV |
1913 | /* "err" is actually length in this case */ |
1914 | err = move_addr_to_user(&address, err, usockaddr, usockaddr_len); | |
1da177e4 LT |
1915 | |
1916 | out_put: | |
6cb153ca | 1917 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1918 | out: |
1919 | return err; | |
1920 | } | |
1921 | ||
8882a107 DB |
1922 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1923 | int __user *, usockaddr_len) | |
1924 | { | |
1925 | return __sys_getsockname(fd, usockaddr, usockaddr_len); | |
1926 | } | |
1927 | ||
1da177e4 LT |
1928 | /* |
1929 | * Get the remote address ('name') of a socket object. Move the obtained | |
1930 | * name to user space. | |
1931 | */ | |
1932 | ||
b21c8f83 DB |
1933 | int __sys_getpeername(int fd, struct sockaddr __user *usockaddr, |
1934 | int __user *usockaddr_len) | |
1da177e4 LT |
1935 | { |
1936 | struct socket *sock; | |
230b1839 | 1937 | struct sockaddr_storage address; |
9b2c45d4 | 1938 | int err, fput_needed; |
1da177e4 | 1939 | |
89bddce5 SH |
1940 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1941 | if (sock != NULL) { | |
1da177e4 LT |
1942 | err = security_socket_getpeername(sock); |
1943 | if (err) { | |
6cb153ca | 1944 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1945 | return err; |
1946 | } | |
1947 | ||
9b2c45d4 DV |
1948 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 1); |
1949 | if (err >= 0) | |
1950 | /* "err" is actually length in this case */ | |
1951 | err = move_addr_to_user(&address, err, usockaddr, | |
89bddce5 | 1952 | usockaddr_len); |
6cb153ca | 1953 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1954 | } |
1955 | return err; | |
1956 | } | |
1957 | ||
b21c8f83 DB |
1958 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1959 | int __user *, usockaddr_len) | |
1960 | { | |
1961 | return __sys_getpeername(fd, usockaddr, usockaddr_len); | |
1962 | } | |
1963 | ||
1da177e4 LT |
1964 | /* |
1965 | * Send a datagram to a given address. We move the address into kernel | |
1966 | * space and check the user space data area is readable before invoking | |
1967 | * the protocol. | |
1968 | */ | |
211b634b DB |
1969 | int __sys_sendto(int fd, void __user *buff, size_t len, unsigned int flags, |
1970 | struct sockaddr __user *addr, int addr_len) | |
1da177e4 LT |
1971 | { |
1972 | struct socket *sock; | |
230b1839 | 1973 | struct sockaddr_storage address; |
1da177e4 LT |
1974 | int err; |
1975 | struct msghdr msg; | |
1976 | struct iovec iov; | |
6cb153ca | 1977 | int fput_needed; |
6cb153ca | 1978 | |
602bd0e9 AV |
1979 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1980 | if (unlikely(err)) | |
1981 | return err; | |
de0fa95c PE |
1982 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1983 | if (!sock) | |
4387ff75 | 1984 | goto out; |
6cb153ca | 1985 | |
89bddce5 | 1986 | msg.msg_name = NULL; |
89bddce5 SH |
1987 | msg.msg_control = NULL; |
1988 | msg.msg_controllen = 0; | |
1989 | msg.msg_namelen = 0; | |
6cb153ca | 1990 | if (addr) { |
43db362d | 1991 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1992 | if (err < 0) |
1993 | goto out_put; | |
230b1839 | 1994 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1995 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1996 | } |
1997 | if (sock->file->f_flags & O_NONBLOCK) | |
1998 | flags |= MSG_DONTWAIT; | |
1999 | msg.msg_flags = flags; | |
d8725c86 | 2000 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 2001 | |
89bddce5 | 2002 | out_put: |
de0fa95c | 2003 | fput_light(sock->file, fput_needed); |
4387ff75 | 2004 | out: |
1da177e4 LT |
2005 | return err; |
2006 | } | |
2007 | ||
211b634b DB |
2008 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
2009 | unsigned int, flags, struct sockaddr __user *, addr, | |
2010 | int, addr_len) | |
2011 | { | |
2012 | return __sys_sendto(fd, buff, len, flags, addr, addr_len); | |
2013 | } | |
2014 | ||
1da177e4 | 2015 | /* |
89bddce5 | 2016 | * Send a datagram down a socket. |
1da177e4 LT |
2017 | */ |
2018 | ||
3e0fa65f | 2019 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 2020 | unsigned int, flags) |
1da177e4 | 2021 | { |
211b634b | 2022 | return __sys_sendto(fd, buff, len, flags, NULL, 0); |
1da177e4 LT |
2023 | } |
2024 | ||
2025 | /* | |
89bddce5 | 2026 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
2027 | * sender. We verify the buffers are writable and if needed move the |
2028 | * sender address from kernel to user space. | |
2029 | */ | |
7a09e1eb DB |
2030 | int __sys_recvfrom(int fd, void __user *ubuf, size_t size, unsigned int flags, |
2031 | struct sockaddr __user *addr, int __user *addr_len) | |
1da177e4 LT |
2032 | { |
2033 | struct socket *sock; | |
2034 | struct iovec iov; | |
2035 | struct msghdr msg; | |
230b1839 | 2036 | struct sockaddr_storage address; |
89bddce5 | 2037 | int err, err2; |
6cb153ca BL |
2038 | int fput_needed; |
2039 | ||
602bd0e9 AV |
2040 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
2041 | if (unlikely(err)) | |
2042 | return err; | |
de0fa95c | 2043 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 2044 | if (!sock) |
de0fa95c | 2045 | goto out; |
1da177e4 | 2046 | |
89bddce5 SH |
2047 | msg.msg_control = NULL; |
2048 | msg.msg_controllen = 0; | |
f3d33426 HFS |
2049 | /* Save some cycles and don't copy the address if not needed */ |
2050 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
2051 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
2052 | msg.msg_namelen = 0; | |
130ed5d1 | 2053 | msg.msg_iocb = NULL; |
9f138fa6 | 2054 | msg.msg_flags = 0; |
1da177e4 LT |
2055 | if (sock->file->f_flags & O_NONBLOCK) |
2056 | flags |= MSG_DONTWAIT; | |
2da62906 | 2057 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 2058 | |
89bddce5 | 2059 | if (err >= 0 && addr != NULL) { |
43db362d | 2060 | err2 = move_addr_to_user(&address, |
230b1839 | 2061 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
2062 | if (err2 < 0) |
2063 | err = err2; | |
1da177e4 | 2064 | } |
de0fa95c PE |
2065 | |
2066 | fput_light(sock->file, fput_needed); | |
4387ff75 | 2067 | out: |
1da177e4 LT |
2068 | return err; |
2069 | } | |
2070 | ||
7a09e1eb DB |
2071 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
2072 | unsigned int, flags, struct sockaddr __user *, addr, | |
2073 | int __user *, addr_len) | |
2074 | { | |
2075 | return __sys_recvfrom(fd, ubuf, size, flags, addr, addr_len); | |
2076 | } | |
2077 | ||
1da177e4 | 2078 | /* |
89bddce5 | 2079 | * Receive a datagram from a socket. |
1da177e4 LT |
2080 | */ |
2081 | ||
b7c0ddf5 JG |
2082 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
2083 | unsigned int, flags) | |
1da177e4 | 2084 | { |
7a09e1eb | 2085 | return __sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
1da177e4 LT |
2086 | } |
2087 | ||
2088 | /* | |
2089 | * Set a socket option. Because we don't know the option lengths we have | |
2090 | * to pass the user mode parameter for the protocols to sort out. | |
2091 | */ | |
2092 | ||
cc36dca0 DB |
2093 | static int __sys_setsockopt(int fd, int level, int optname, |
2094 | char __user *optval, int optlen) | |
1da177e4 | 2095 | { |
0d01da6a SF |
2096 | mm_segment_t oldfs = get_fs(); |
2097 | char *kernel_optval = NULL; | |
6cb153ca | 2098 | int err, fput_needed; |
1da177e4 LT |
2099 | struct socket *sock; |
2100 | ||
2101 | if (optlen < 0) | |
2102 | return -EINVAL; | |
89bddce5 SH |
2103 | |
2104 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2105 | if (sock != NULL) { | |
2106 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
2107 | if (err) |
2108 | goto out_put; | |
1da177e4 | 2109 | |
0d01da6a SF |
2110 | err = BPF_CGROUP_RUN_PROG_SETSOCKOPT(sock->sk, &level, |
2111 | &optname, optval, &optlen, | |
2112 | &kernel_optval); | |
2113 | ||
2114 | if (err < 0) { | |
2115 | goto out_put; | |
2116 | } else if (err > 0) { | |
2117 | err = 0; | |
2118 | goto out_put; | |
2119 | } | |
2120 | ||
2121 | if (kernel_optval) { | |
2122 | set_fs(KERNEL_DS); | |
2123 | optval = (char __user __force *)kernel_optval; | |
2124 | } | |
2125 | ||
1da177e4 | 2126 | if (level == SOL_SOCKET) |
89bddce5 SH |
2127 | err = |
2128 | sock_setsockopt(sock, level, optname, optval, | |
2129 | optlen); | |
1da177e4 | 2130 | else |
89bddce5 SH |
2131 | err = |
2132 | sock->ops->setsockopt(sock, level, optname, optval, | |
2133 | optlen); | |
0d01da6a SF |
2134 | |
2135 | if (kernel_optval) { | |
2136 | set_fs(oldfs); | |
2137 | kfree(kernel_optval); | |
2138 | } | |
6cb153ca BL |
2139 | out_put: |
2140 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2141 | } |
2142 | return err; | |
2143 | } | |
2144 | ||
cc36dca0 DB |
2145 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
2146 | char __user *, optval, int, optlen) | |
2147 | { | |
2148 | return __sys_setsockopt(fd, level, optname, optval, optlen); | |
2149 | } | |
2150 | ||
1da177e4 LT |
2151 | /* |
2152 | * Get a socket option. Because we don't know the option lengths we have | |
2153 | * to pass a user mode parameter for the protocols to sort out. | |
2154 | */ | |
2155 | ||
13a2d70e DB |
2156 | static int __sys_getsockopt(int fd, int level, int optname, |
2157 | char __user *optval, int __user *optlen) | |
1da177e4 | 2158 | { |
6cb153ca | 2159 | int err, fput_needed; |
1da177e4 | 2160 | struct socket *sock; |
0d01da6a | 2161 | int max_optlen; |
1da177e4 | 2162 | |
89bddce5 SH |
2163 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2164 | if (sock != NULL) { | |
6cb153ca BL |
2165 | err = security_socket_getsockopt(sock, level, optname); |
2166 | if (err) | |
2167 | goto out_put; | |
1da177e4 | 2168 | |
0d01da6a SF |
2169 | max_optlen = BPF_CGROUP_GETSOCKOPT_MAX_OPTLEN(optlen); |
2170 | ||
1da177e4 | 2171 | if (level == SOL_SOCKET) |
89bddce5 SH |
2172 | err = |
2173 | sock_getsockopt(sock, level, optname, optval, | |
2174 | optlen); | |
1da177e4 | 2175 | else |
89bddce5 SH |
2176 | err = |
2177 | sock->ops->getsockopt(sock, level, optname, optval, | |
2178 | optlen); | |
0d01da6a SF |
2179 | |
2180 | err = BPF_CGROUP_RUN_PROG_GETSOCKOPT(sock->sk, level, optname, | |
2181 | optval, optlen, | |
2182 | max_optlen, err); | |
6cb153ca BL |
2183 | out_put: |
2184 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2185 | } |
2186 | return err; | |
2187 | } | |
2188 | ||
13a2d70e DB |
2189 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
2190 | char __user *, optval, int __user *, optlen) | |
2191 | { | |
2192 | return __sys_getsockopt(fd, level, optname, optval, optlen); | |
2193 | } | |
2194 | ||
1da177e4 LT |
2195 | /* |
2196 | * Shutdown a socket. | |
2197 | */ | |
2198 | ||
005a1aea | 2199 | int __sys_shutdown(int fd, int how) |
1da177e4 | 2200 | { |
6cb153ca | 2201 | int err, fput_needed; |
1da177e4 LT |
2202 | struct socket *sock; |
2203 | ||
89bddce5 SH |
2204 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2205 | if (sock != NULL) { | |
1da177e4 | 2206 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
2207 | if (!err) |
2208 | err = sock->ops->shutdown(sock, how); | |
2209 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2210 | } |
2211 | return err; | |
2212 | } | |
2213 | ||
005a1aea DB |
2214 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
2215 | { | |
2216 | return __sys_shutdown(fd, how); | |
2217 | } | |
2218 | ||
89bddce5 | 2219 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
2220 | * fields which are the same type (int / unsigned) on our platforms. |
2221 | */ | |
2222 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
2223 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
2224 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
2225 | ||
c71d8ebe TH |
2226 | struct used_address { |
2227 | struct sockaddr_storage name; | |
2228 | unsigned int name_len; | |
2229 | }; | |
2230 | ||
0a384abf JA |
2231 | int __copy_msghdr_from_user(struct msghdr *kmsg, |
2232 | struct user_msghdr __user *umsg, | |
2233 | struct sockaddr __user **save_addr, | |
2234 | struct iovec __user **uiov, size_t *nsegs) | |
1661bf36 | 2235 | { |
ffb07550 | 2236 | struct user_msghdr msg; |
08adb7da AV |
2237 | ssize_t err; |
2238 | ||
ffb07550 | 2239 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 2240 | return -EFAULT; |
dbb490b9 | 2241 | |
864d9664 | 2242 | kmsg->msg_control = (void __force *)msg.msg_control; |
ffb07550 AV |
2243 | kmsg->msg_controllen = msg.msg_controllen; |
2244 | kmsg->msg_flags = msg.msg_flags; | |
2245 | ||
2246 | kmsg->msg_namelen = msg.msg_namelen; | |
2247 | if (!msg.msg_name) | |
6a2a2b3a AS |
2248 | kmsg->msg_namelen = 0; |
2249 | ||
dbb490b9 ML |
2250 | if (kmsg->msg_namelen < 0) |
2251 | return -EINVAL; | |
2252 | ||
1661bf36 | 2253 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 2254 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
2255 | |
2256 | if (save_addr) | |
ffb07550 | 2257 | *save_addr = msg.msg_name; |
08adb7da | 2258 | |
ffb07550 | 2259 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 2260 | if (!save_addr) { |
864d9664 PA |
2261 | err = move_addr_to_kernel(msg.msg_name, |
2262 | kmsg->msg_namelen, | |
08adb7da AV |
2263 | kmsg->msg_name); |
2264 | if (err < 0) | |
2265 | return err; | |
2266 | } | |
2267 | } else { | |
2268 | kmsg->msg_name = NULL; | |
2269 | kmsg->msg_namelen = 0; | |
2270 | } | |
2271 | ||
ffb07550 | 2272 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
2273 | return -EMSGSIZE; |
2274 | ||
0345f931 | 2275 | kmsg->msg_iocb = NULL; |
0a384abf JA |
2276 | *uiov = msg.msg_iov; |
2277 | *nsegs = msg.msg_iovlen; | |
2278 | return 0; | |
2279 | } | |
2280 | ||
2281 | static int copy_msghdr_from_user(struct msghdr *kmsg, | |
2282 | struct user_msghdr __user *umsg, | |
2283 | struct sockaddr __user **save_addr, | |
2284 | struct iovec **iov) | |
2285 | { | |
2286 | struct user_msghdr msg; | |
2287 | ssize_t err; | |
2288 | ||
2289 | err = __copy_msghdr_from_user(kmsg, umsg, save_addr, &msg.msg_iov, | |
2290 | &msg.msg_iovlen); | |
2291 | if (err) | |
2292 | return err; | |
0345f931 | 2293 | |
87e5e6da | 2294 | err = import_iovec(save_addr ? READ : WRITE, |
ffb07550 | 2295 | msg.msg_iov, msg.msg_iovlen, |
da184284 | 2296 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
87e5e6da | 2297 | return err < 0 ? err : 0; |
1661bf36 DC |
2298 | } |
2299 | ||
4257c8ca JA |
2300 | static int ____sys_sendmsg(struct socket *sock, struct msghdr *msg_sys, |
2301 | unsigned int flags, struct used_address *used_address, | |
2302 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 2303 | { |
b9d717a7 | 2304 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 2305 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 2306 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 2307 | unsigned char *ctl_buf = ctl; |
d8725c86 | 2308 | int ctl_len; |
08adb7da | 2309 | ssize_t err; |
89bddce5 | 2310 | |
1da177e4 LT |
2311 | err = -ENOBUFS; |
2312 | ||
228e548e | 2313 | if (msg_sys->msg_controllen > INT_MAX) |
4257c8ca | 2314 | goto out; |
28a94d8f | 2315 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2316 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2317 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2318 | err = |
228e548e | 2319 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2320 | sizeof(ctl)); |
1da177e4 | 2321 | if (err) |
4257c8ca | 2322 | goto out; |
228e548e AB |
2323 | ctl_buf = msg_sys->msg_control; |
2324 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2325 | } else if (ctl_len) { |
ac4340fc DM |
2326 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2327 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2328 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2329 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2330 | if (ctl_buf == NULL) |
4257c8ca | 2331 | goto out; |
1da177e4 LT |
2332 | } |
2333 | err = -EFAULT; | |
2334 | /* | |
228e548e | 2335 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2336 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2337 | * checking falls down on this. | |
2338 | */ | |
fb8621bb | 2339 | if (copy_from_user(ctl_buf, |
228e548e | 2340 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2341 | ctl_len)) |
1da177e4 | 2342 | goto out_freectl; |
228e548e | 2343 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2344 | } |
228e548e | 2345 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2346 | |
2347 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2348 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2349 | /* |
2350 | * If this is sendmmsg() and current destination address is same as | |
2351 | * previously succeeded address, omit asking LSM's decision. | |
2352 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2353 | * destination address never matches. | |
2354 | */ | |
bc909d9d MD |
2355 | if (used_address && msg_sys->msg_name && |
2356 | used_address->name_len == msg_sys->msg_namelen && | |
2357 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2358 | used_address->name_len)) { |
d8725c86 | 2359 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2360 | goto out_freectl; |
2361 | } | |
d8725c86 | 2362 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2363 | /* |
2364 | * If this is sendmmsg() and sending to current destination address was | |
2365 | * successful, remember it. | |
2366 | */ | |
2367 | if (used_address && err >= 0) { | |
2368 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2369 | if (msg_sys->msg_name) |
2370 | memcpy(&used_address->name, msg_sys->msg_name, | |
2371 | used_address->name_len); | |
c71d8ebe | 2372 | } |
1da177e4 LT |
2373 | |
2374 | out_freectl: | |
89bddce5 | 2375 | if (ctl_buf != ctl) |
1da177e4 | 2376 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
4257c8ca JA |
2377 | out: |
2378 | return err; | |
2379 | } | |
2380 | ||
03b1230c JA |
2381 | int sendmsg_copy_msghdr(struct msghdr *msg, |
2382 | struct user_msghdr __user *umsg, unsigned flags, | |
2383 | struct iovec **iov) | |
4257c8ca JA |
2384 | { |
2385 | int err; | |
2386 | ||
2387 | if (flags & MSG_CMSG_COMPAT) { | |
2388 | struct compat_msghdr __user *msg_compat; | |
2389 | ||
2390 | msg_compat = (struct compat_msghdr __user *) umsg; | |
2391 | err = get_compat_msghdr(msg, msg_compat, NULL, iov); | |
2392 | } else { | |
2393 | err = copy_msghdr_from_user(msg, umsg, NULL, iov); | |
2394 | } | |
2395 | if (err < 0) | |
2396 | return err; | |
2397 | ||
2398 | return 0; | |
2399 | } | |
2400 | ||
2401 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, | |
2402 | struct msghdr *msg_sys, unsigned int flags, | |
2403 | struct used_address *used_address, | |
2404 | unsigned int allowed_msghdr_flags) | |
2405 | { | |
2406 | struct sockaddr_storage address; | |
2407 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; | |
2408 | ssize_t err; | |
2409 | ||
2410 | msg_sys->msg_name = &address; | |
2411 | ||
2412 | err = sendmsg_copy_msghdr(msg_sys, msg, flags, &iov); | |
2413 | if (err < 0) | |
2414 | return err; | |
2415 | ||
2416 | err = ____sys_sendmsg(sock, msg_sys, flags, used_address, | |
2417 | allowed_msghdr_flags); | |
da184284 | 2418 | kfree(iov); |
228e548e AB |
2419 | return err; |
2420 | } | |
2421 | ||
2422 | /* | |
2423 | * BSD sendmsg interface | |
2424 | */ | |
03b1230c | 2425 | long __sys_sendmsg_sock(struct socket *sock, struct msghdr *msg, |
0fa03c62 JA |
2426 | unsigned int flags) |
2427 | { | |
d69e0779 | 2428 | /* disallow ancillary data requests from this path */ |
03b1230c JA |
2429 | if (msg->msg_control || msg->msg_controllen) |
2430 | return -EINVAL; | |
d69e0779 | 2431 | |
03b1230c | 2432 | return ____sys_sendmsg(sock, msg, flags, NULL, 0); |
0fa03c62 | 2433 | } |
228e548e | 2434 | |
e1834a32 DB |
2435 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2436 | bool forbid_cmsg_compat) | |
228e548e AB |
2437 | { |
2438 | int fput_needed, err; | |
2439 | struct msghdr msg_sys; | |
1be374a0 AL |
2440 | struct socket *sock; |
2441 | ||
e1834a32 DB |
2442 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2443 | return -EINVAL; | |
2444 | ||
1be374a0 | 2445 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2446 | if (!sock) |
2447 | goto out; | |
2448 | ||
28a94d8f | 2449 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2450 | |
6cb153ca | 2451 | fput_light(sock->file, fput_needed); |
89bddce5 | 2452 | out: |
1da177e4 LT |
2453 | return err; |
2454 | } | |
2455 | ||
666547ff | 2456 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 | 2457 | { |
e1834a32 | 2458 | return __sys_sendmsg(fd, msg, flags, true); |
a7526eb5 AL |
2459 | } |
2460 | ||
228e548e AB |
2461 | /* |
2462 | * Linux sendmmsg interface | |
2463 | */ | |
2464 | ||
2465 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
e1834a32 | 2466 | unsigned int flags, bool forbid_cmsg_compat) |
228e548e AB |
2467 | { |
2468 | int fput_needed, err, datagrams; | |
2469 | struct socket *sock; | |
2470 | struct mmsghdr __user *entry; | |
2471 | struct compat_mmsghdr __user *compat_entry; | |
2472 | struct msghdr msg_sys; | |
c71d8ebe | 2473 | struct used_address used_address; |
f092276d | 2474 | unsigned int oflags = flags; |
228e548e | 2475 | |
e1834a32 DB |
2476 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2477 | return -EINVAL; | |
2478 | ||
98382f41 AB |
2479 | if (vlen > UIO_MAXIOV) |
2480 | vlen = UIO_MAXIOV; | |
228e548e AB |
2481 | |
2482 | datagrams = 0; | |
2483 | ||
2484 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2485 | if (!sock) | |
2486 | return err; | |
2487 | ||
c71d8ebe | 2488 | used_address.name_len = UINT_MAX; |
228e548e AB |
2489 | entry = mmsg; |
2490 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2491 | err = 0; |
f092276d | 2492 | flags |= MSG_BATCH; |
228e548e AB |
2493 | |
2494 | while (datagrams < vlen) { | |
f092276d TH |
2495 | if (datagrams == vlen - 1) |
2496 | flags = oflags; | |
2497 | ||
228e548e | 2498 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2499 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2500 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2501 | if (err < 0) |
2502 | break; | |
2503 | err = __put_user(err, &compat_entry->msg_len); | |
2504 | ++compat_entry; | |
2505 | } else { | |
a7526eb5 | 2506 | err = ___sys_sendmsg(sock, |
666547ff | 2507 | (struct user_msghdr __user *)entry, |
28a94d8f | 2508 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2509 | if (err < 0) |
2510 | break; | |
2511 | err = put_user(err, &entry->msg_len); | |
2512 | ++entry; | |
2513 | } | |
2514 | ||
2515 | if (err) | |
2516 | break; | |
2517 | ++datagrams; | |
3023898b SHY |
2518 | if (msg_data_left(&msg_sys)) |
2519 | break; | |
a78cb84c | 2520 | cond_resched(); |
228e548e AB |
2521 | } |
2522 | ||
228e548e AB |
2523 | fput_light(sock->file, fput_needed); |
2524 | ||
728ffb86 AB |
2525 | /* We only return an error if no datagrams were able to be sent */ |
2526 | if (datagrams != 0) | |
228e548e AB |
2527 | return datagrams; |
2528 | ||
228e548e AB |
2529 | return err; |
2530 | } | |
2531 | ||
2532 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2533 | unsigned int, vlen, unsigned int, flags) | |
2534 | { | |
e1834a32 | 2535 | return __sys_sendmmsg(fd, mmsg, vlen, flags, true); |
228e548e AB |
2536 | } |
2537 | ||
03b1230c JA |
2538 | int recvmsg_copy_msghdr(struct msghdr *msg, |
2539 | struct user_msghdr __user *umsg, unsigned flags, | |
2540 | struct sockaddr __user **uaddr, | |
2541 | struct iovec **iov) | |
1da177e4 | 2542 | { |
08adb7da | 2543 | ssize_t err; |
1da177e4 | 2544 | |
4257c8ca JA |
2545 | if (MSG_CMSG_COMPAT & flags) { |
2546 | struct compat_msghdr __user *msg_compat; | |
1da177e4 | 2547 | |
4257c8ca JA |
2548 | msg_compat = (struct compat_msghdr __user *) umsg; |
2549 | err = get_compat_msghdr(msg, msg_compat, uaddr, iov); | |
2550 | } else { | |
2551 | err = copy_msghdr_from_user(msg, umsg, uaddr, iov); | |
2552 | } | |
1da177e4 | 2553 | if (err < 0) |
da184284 | 2554 | return err; |
1da177e4 | 2555 | |
4257c8ca JA |
2556 | return 0; |
2557 | } | |
2558 | ||
2559 | static int ____sys_recvmsg(struct socket *sock, struct msghdr *msg_sys, | |
2560 | struct user_msghdr __user *msg, | |
2561 | struct sockaddr __user *uaddr, | |
2562 | unsigned int flags, int nosec) | |
2563 | { | |
2564 | struct compat_msghdr __user *msg_compat = | |
2565 | (struct compat_msghdr __user *) msg; | |
2566 | int __user *uaddr_len = COMPAT_NAMELEN(msg); | |
2567 | struct sockaddr_storage addr; | |
2568 | unsigned long cmsg_ptr; | |
2569 | int len; | |
2570 | ssize_t err; | |
2571 | ||
2572 | msg_sys->msg_name = &addr; | |
a2e27255 ACM |
2573 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2574 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2575 | |
f3d33426 HFS |
2576 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2577 | msg_sys->msg_namelen = 0; | |
2578 | ||
1da177e4 LT |
2579 | if (sock->file->f_flags & O_NONBLOCK) |
2580 | flags |= MSG_DONTWAIT; | |
1af66221 ED |
2581 | |
2582 | if (unlikely(nosec)) | |
2583 | err = sock_recvmsg_nosec(sock, msg_sys, flags); | |
2584 | else | |
2585 | err = sock_recvmsg(sock, msg_sys, flags); | |
2586 | ||
1da177e4 | 2587 | if (err < 0) |
4257c8ca | 2588 | goto out; |
1da177e4 LT |
2589 | len = err; |
2590 | ||
2591 | if (uaddr != NULL) { | |
43db362d | 2592 | err = move_addr_to_user(&addr, |
a2e27255 | 2593 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2594 | uaddr_len); |
1da177e4 | 2595 | if (err < 0) |
4257c8ca | 2596 | goto out; |
1da177e4 | 2597 | } |
a2e27255 | 2598 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2599 | COMPAT_FLAGS(msg)); |
1da177e4 | 2600 | if (err) |
4257c8ca | 2601 | goto out; |
1da177e4 | 2602 | if (MSG_CMSG_COMPAT & flags) |
a2e27255 | 2603 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2604 | &msg_compat->msg_controllen); |
2605 | else | |
a2e27255 | 2606 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2607 | &msg->msg_controllen); |
2608 | if (err) | |
4257c8ca | 2609 | goto out; |
1da177e4 | 2610 | err = len; |
4257c8ca JA |
2611 | out: |
2612 | return err; | |
2613 | } | |
2614 | ||
2615 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, | |
2616 | struct msghdr *msg_sys, unsigned int flags, int nosec) | |
2617 | { | |
2618 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; | |
2619 | /* user mode address pointers */ | |
2620 | struct sockaddr __user *uaddr; | |
2621 | ssize_t err; | |
2622 | ||
2623 | err = recvmsg_copy_msghdr(msg_sys, msg, flags, &uaddr, &iov); | |
2624 | if (err < 0) | |
2625 | return err; | |
1da177e4 | 2626 | |
4257c8ca | 2627 | err = ____sys_recvmsg(sock, msg_sys, msg, uaddr, flags, nosec); |
da184284 | 2628 | kfree(iov); |
a2e27255 ACM |
2629 | return err; |
2630 | } | |
2631 | ||
2632 | /* | |
2633 | * BSD recvmsg interface | |
2634 | */ | |
2635 | ||
03b1230c JA |
2636 | long __sys_recvmsg_sock(struct socket *sock, struct msghdr *msg, |
2637 | struct user_msghdr __user *umsg, | |
2638 | struct sockaddr __user *uaddr, unsigned int flags) | |
aa1fa28f | 2639 | { |
d69e0779 | 2640 | /* disallow ancillary data requests from this path */ |
03b1230c JA |
2641 | if (msg->msg_control || msg->msg_controllen) |
2642 | return -EINVAL; | |
aa1fa28f | 2643 | |
03b1230c | 2644 | return ____sys_recvmsg(sock, msg, umsg, uaddr, flags, 0); |
aa1fa28f JA |
2645 | } |
2646 | ||
e1834a32 DB |
2647 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2648 | bool forbid_cmsg_compat) | |
a2e27255 ACM |
2649 | { |
2650 | int fput_needed, err; | |
2651 | struct msghdr msg_sys; | |
1be374a0 AL |
2652 | struct socket *sock; |
2653 | ||
e1834a32 DB |
2654 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2655 | return -EINVAL; | |
2656 | ||
1be374a0 | 2657 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2658 | if (!sock) |
2659 | goto out; | |
2660 | ||
a7526eb5 | 2661 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2662 | |
6cb153ca | 2663 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2664 | out: |
2665 | return err; | |
2666 | } | |
2667 | ||
666547ff | 2668 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2669 | unsigned int, flags) |
2670 | { | |
e1834a32 | 2671 | return __sys_recvmsg(fd, msg, flags, true); |
a7526eb5 AL |
2672 | } |
2673 | ||
a2e27255 ACM |
2674 | /* |
2675 | * Linux recvmmsg interface | |
2676 | */ | |
2677 | ||
e11d4284 AB |
2678 | static int do_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2679 | unsigned int vlen, unsigned int flags, | |
2680 | struct timespec64 *timeout) | |
a2e27255 ACM |
2681 | { |
2682 | int fput_needed, err, datagrams; | |
2683 | struct socket *sock; | |
2684 | struct mmsghdr __user *entry; | |
d7256d0e | 2685 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2686 | struct msghdr msg_sys; |
766b9f92 DD |
2687 | struct timespec64 end_time; |
2688 | struct timespec64 timeout64; | |
a2e27255 ACM |
2689 | |
2690 | if (timeout && | |
2691 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2692 | timeout->tv_nsec)) | |
2693 | return -EINVAL; | |
2694 | ||
2695 | datagrams = 0; | |
2696 | ||
2697 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2698 | if (!sock) | |
2699 | return err; | |
2700 | ||
7797dc41 SHY |
2701 | if (likely(!(flags & MSG_ERRQUEUE))) { |
2702 | err = sock_error(sock->sk); | |
2703 | if (err) { | |
2704 | datagrams = err; | |
2705 | goto out_put; | |
2706 | } | |
e623a9e9 | 2707 | } |
a2e27255 ACM |
2708 | |
2709 | entry = mmsg; | |
d7256d0e | 2710 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2711 | |
2712 | while (datagrams < vlen) { | |
2713 | /* | |
2714 | * No need to ask LSM for more than the first datagram. | |
2715 | */ | |
d7256d0e | 2716 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2717 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2718 | &msg_sys, flags & ~MSG_WAITFORONE, |
2719 | datagrams); | |
d7256d0e JMG |
2720 | if (err < 0) |
2721 | break; | |
2722 | err = __put_user(err, &compat_entry->msg_len); | |
2723 | ++compat_entry; | |
2724 | } else { | |
a7526eb5 | 2725 | err = ___sys_recvmsg(sock, |
666547ff | 2726 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2727 | &msg_sys, flags & ~MSG_WAITFORONE, |
2728 | datagrams); | |
d7256d0e JMG |
2729 | if (err < 0) |
2730 | break; | |
2731 | err = put_user(err, &entry->msg_len); | |
2732 | ++entry; | |
2733 | } | |
2734 | ||
a2e27255 ACM |
2735 | if (err) |
2736 | break; | |
a2e27255 ACM |
2737 | ++datagrams; |
2738 | ||
71c5c159 BB |
2739 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2740 | if (flags & MSG_WAITFORONE) | |
2741 | flags |= MSG_DONTWAIT; | |
2742 | ||
a2e27255 | 2743 | if (timeout) { |
766b9f92 | 2744 | ktime_get_ts64(&timeout64); |
c2e6c856 | 2745 | *timeout = timespec64_sub(end_time, timeout64); |
a2e27255 ACM |
2746 | if (timeout->tv_sec < 0) { |
2747 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2748 | break; | |
2749 | } | |
2750 | ||
2751 | /* Timeout, return less than vlen datagrams */ | |
2752 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2753 | break; | |
2754 | } | |
2755 | ||
2756 | /* Out of band data, return right away */ | |
2757 | if (msg_sys.msg_flags & MSG_OOB) | |
2758 | break; | |
a78cb84c | 2759 | cond_resched(); |
a2e27255 ACM |
2760 | } |
2761 | ||
a2e27255 | 2762 | if (err == 0) |
34b88a68 ACM |
2763 | goto out_put; |
2764 | ||
2765 | if (datagrams == 0) { | |
2766 | datagrams = err; | |
2767 | goto out_put; | |
2768 | } | |
a2e27255 | 2769 | |
34b88a68 ACM |
2770 | /* |
2771 | * We may return less entries than requested (vlen) if the | |
2772 | * sock is non block and there aren't enough datagrams... | |
2773 | */ | |
2774 | if (err != -EAGAIN) { | |
a2e27255 | 2775 | /* |
34b88a68 ACM |
2776 | * ... or if recvmsg returns an error after we |
2777 | * received some datagrams, where we record the | |
2778 | * error to return on the next call or if the | |
2779 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2780 | */ |
34b88a68 | 2781 | sock->sk->sk_err = -err; |
a2e27255 | 2782 | } |
34b88a68 ACM |
2783 | out_put: |
2784 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2785 | |
34b88a68 | 2786 | return datagrams; |
a2e27255 ACM |
2787 | } |
2788 | ||
e11d4284 AB |
2789 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2790 | unsigned int vlen, unsigned int flags, | |
2791 | struct __kernel_timespec __user *timeout, | |
2792 | struct old_timespec32 __user *timeout32) | |
a2e27255 ACM |
2793 | { |
2794 | int datagrams; | |
c2e6c856 | 2795 | struct timespec64 timeout_sys; |
a2e27255 | 2796 | |
e11d4284 AB |
2797 | if (timeout && get_timespec64(&timeout_sys, timeout)) |
2798 | return -EFAULT; | |
a2e27255 | 2799 | |
e11d4284 | 2800 | if (timeout32 && get_old_timespec32(&timeout_sys, timeout32)) |
a2e27255 ACM |
2801 | return -EFAULT; |
2802 | ||
e11d4284 AB |
2803 | if (!timeout && !timeout32) |
2804 | return do_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2805 | ||
2806 | datagrams = do_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
a2e27255 | 2807 | |
e11d4284 AB |
2808 | if (datagrams <= 0) |
2809 | return datagrams; | |
2810 | ||
2811 | if (timeout && put_timespec64(&timeout_sys, timeout)) | |
2812 | datagrams = -EFAULT; | |
2813 | ||
2814 | if (timeout32 && put_old_timespec32(&timeout_sys, timeout32)) | |
a2e27255 ACM |
2815 | datagrams = -EFAULT; |
2816 | ||
2817 | return datagrams; | |
2818 | } | |
2819 | ||
1255e269 DB |
2820 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, |
2821 | unsigned int, vlen, unsigned int, flags, | |
c2e6c856 | 2822 | struct __kernel_timespec __user *, timeout) |
1255e269 | 2823 | { |
e11d4284 AB |
2824 | if (flags & MSG_CMSG_COMPAT) |
2825 | return -EINVAL; | |
2826 | ||
2827 | return __sys_recvmmsg(fd, mmsg, vlen, flags, timeout, NULL); | |
2828 | } | |
2829 | ||
2830 | #ifdef CONFIG_COMPAT_32BIT_TIME | |
2831 | SYSCALL_DEFINE5(recvmmsg_time32, int, fd, struct mmsghdr __user *, mmsg, | |
2832 | unsigned int, vlen, unsigned int, flags, | |
2833 | struct old_timespec32 __user *, timeout) | |
2834 | { | |
2835 | if (flags & MSG_CMSG_COMPAT) | |
2836 | return -EINVAL; | |
2837 | ||
2838 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL, timeout); | |
1255e269 | 2839 | } |
e11d4284 | 2840 | #endif |
1255e269 | 2841 | |
a2e27255 | 2842 | #ifdef __ARCH_WANT_SYS_SOCKETCALL |
1da177e4 LT |
2843 | /* Argument list sizes for sys_socketcall */ |
2844 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2845 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2846 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2847 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2848 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2849 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2850 | }; |
2851 | ||
1da177e4 LT |
2852 | #undef AL |
2853 | ||
2854 | /* | |
89bddce5 | 2855 | * System call vectors. |
1da177e4 LT |
2856 | * |
2857 | * Argument checking cleaned up. Saved 20% in size. | |
2858 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2859 | * it is set by the callees. |
1da177e4 LT |
2860 | */ |
2861 | ||
3e0fa65f | 2862 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2863 | { |
2950fa9d | 2864 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2865 | unsigned long a0, a1; |
1da177e4 | 2866 | int err; |
47379052 | 2867 | unsigned int len; |
1da177e4 | 2868 | |
228e548e | 2869 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 | 2870 | return -EINVAL; |
c8e8cd57 | 2871 | call = array_index_nospec(call, SYS_SENDMMSG + 1); |
1da177e4 | 2872 | |
47379052 AV |
2873 | len = nargs[call]; |
2874 | if (len > sizeof(a)) | |
2875 | return -EINVAL; | |
2876 | ||
1da177e4 | 2877 | /* copy_from_user should be SMP safe. */ |
47379052 | 2878 | if (copy_from_user(a, args, len)) |
1da177e4 | 2879 | return -EFAULT; |
3ec3b2fb | 2880 | |
2950fa9d CG |
2881 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2882 | if (err) | |
2883 | return err; | |
3ec3b2fb | 2884 | |
89bddce5 SH |
2885 | a0 = a[0]; |
2886 | a1 = a[1]; | |
2887 | ||
2888 | switch (call) { | |
2889 | case SYS_SOCKET: | |
9d6a15c3 | 2890 | err = __sys_socket(a0, a1, a[2]); |
89bddce5 SH |
2891 | break; |
2892 | case SYS_BIND: | |
a87d35d8 | 2893 | err = __sys_bind(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2894 | break; |
2895 | case SYS_CONNECT: | |
1387c2c2 | 2896 | err = __sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2897 | break; |
2898 | case SYS_LISTEN: | |
25e290ee | 2899 | err = __sys_listen(a0, a1); |
89bddce5 SH |
2900 | break; |
2901 | case SYS_ACCEPT: | |
4541e805 DB |
2902 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2903 | (int __user *)a[2], 0); | |
89bddce5 SH |
2904 | break; |
2905 | case SYS_GETSOCKNAME: | |
2906 | err = | |
8882a107 DB |
2907 | __sys_getsockname(a0, (struct sockaddr __user *)a1, |
2908 | (int __user *)a[2]); | |
89bddce5 SH |
2909 | break; |
2910 | case SYS_GETPEERNAME: | |
2911 | err = | |
b21c8f83 DB |
2912 | __sys_getpeername(a0, (struct sockaddr __user *)a1, |
2913 | (int __user *)a[2]); | |
89bddce5 SH |
2914 | break; |
2915 | case SYS_SOCKETPAIR: | |
6debc8d8 | 2916 | err = __sys_socketpair(a0, a1, a[2], (int __user *)a[3]); |
89bddce5 SH |
2917 | break; |
2918 | case SYS_SEND: | |
f3bf896b DB |
2919 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2920 | NULL, 0); | |
89bddce5 SH |
2921 | break; |
2922 | case SYS_SENDTO: | |
211b634b DB |
2923 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2924 | (struct sockaddr __user *)a[4], a[5]); | |
89bddce5 SH |
2925 | break; |
2926 | case SYS_RECV: | |
d27e9afc DB |
2927 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2928 | NULL, NULL); | |
89bddce5 SH |
2929 | break; |
2930 | case SYS_RECVFROM: | |
7a09e1eb DB |
2931 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2932 | (struct sockaddr __user *)a[4], | |
2933 | (int __user *)a[5]); | |
89bddce5 SH |
2934 | break; |
2935 | case SYS_SHUTDOWN: | |
005a1aea | 2936 | err = __sys_shutdown(a0, a1); |
89bddce5 SH |
2937 | break; |
2938 | case SYS_SETSOCKOPT: | |
cc36dca0 DB |
2939 | err = __sys_setsockopt(a0, a1, a[2], (char __user *)a[3], |
2940 | a[4]); | |
89bddce5 SH |
2941 | break; |
2942 | case SYS_GETSOCKOPT: | |
2943 | err = | |
13a2d70e DB |
2944 | __sys_getsockopt(a0, a1, a[2], (char __user *)a[3], |
2945 | (int __user *)a[4]); | |
89bddce5 SH |
2946 | break; |
2947 | case SYS_SENDMSG: | |
e1834a32 DB |
2948 | err = __sys_sendmsg(a0, (struct user_msghdr __user *)a1, |
2949 | a[2], true); | |
89bddce5 | 2950 | break; |
228e548e | 2951 | case SYS_SENDMMSG: |
e1834a32 DB |
2952 | err = __sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
2953 | a[3], true); | |
228e548e | 2954 | break; |
89bddce5 | 2955 | case SYS_RECVMSG: |
e1834a32 DB |
2956 | err = __sys_recvmsg(a0, (struct user_msghdr __user *)a1, |
2957 | a[2], true); | |
89bddce5 | 2958 | break; |
a2e27255 | 2959 | case SYS_RECVMMSG: |
3ca47e95 | 2960 | if (IS_ENABLED(CONFIG_64BIT)) |
e11d4284 AB |
2961 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, |
2962 | a[2], a[3], | |
2963 | (struct __kernel_timespec __user *)a[4], | |
2964 | NULL); | |
2965 | else | |
2966 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, | |
2967 | a[2], a[3], NULL, | |
2968 | (struct old_timespec32 __user *)a[4]); | |
a2e27255 | 2969 | break; |
de11defe | 2970 | case SYS_ACCEPT4: |
4541e805 DB |
2971 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2972 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2973 | break; |
89bddce5 SH |
2974 | default: |
2975 | err = -EINVAL; | |
2976 | break; | |
1da177e4 LT |
2977 | } |
2978 | return err; | |
2979 | } | |
2980 | ||
89bddce5 | 2981 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2982 | |
55737fda SH |
2983 | /** |
2984 | * sock_register - add a socket protocol handler | |
2985 | * @ops: description of protocol | |
2986 | * | |
1da177e4 LT |
2987 | * This function is called by a protocol handler that wants to |
2988 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2989 | * socket interface. The value ops->family corresponds to the |
55737fda | 2990 | * socket system call protocol family. |
1da177e4 | 2991 | */ |
f0fd27d4 | 2992 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2993 | { |
2994 | int err; | |
2995 | ||
2996 | if (ops->family >= NPROTO) { | |
3410f22e | 2997 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2998 | return -ENOBUFS; |
2999 | } | |
55737fda SH |
3000 | |
3001 | spin_lock(&net_family_lock); | |
190683a9 ED |
3002 | if (rcu_dereference_protected(net_families[ops->family], |
3003 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
3004 | err = -EEXIST; |
3005 | else { | |
cf778b00 | 3006 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
3007 | err = 0; |
3008 | } | |
55737fda SH |
3009 | spin_unlock(&net_family_lock); |
3010 | ||
3410f22e | 3011 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
3012 | return err; |
3013 | } | |
c6d409cf | 3014 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 3015 | |
55737fda SH |
3016 | /** |
3017 | * sock_unregister - remove a protocol handler | |
3018 | * @family: protocol family to remove | |
3019 | * | |
1da177e4 LT |
3020 | * This function is called by a protocol handler that wants to |
3021 | * remove its address family, and have it unlinked from the | |
55737fda SH |
3022 | * new socket creation. |
3023 | * | |
3024 | * If protocol handler is a module, then it can use module reference | |
3025 | * counts to protect against new references. If protocol handler is not | |
3026 | * a module then it needs to provide its own protection in | |
3027 | * the ops->create routine. | |
1da177e4 | 3028 | */ |
f0fd27d4 | 3029 | void sock_unregister(int family) |
1da177e4 | 3030 | { |
f0fd27d4 | 3031 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 3032 | |
55737fda | 3033 | spin_lock(&net_family_lock); |
a9b3cd7f | 3034 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
3035 | spin_unlock(&net_family_lock); |
3036 | ||
3037 | synchronize_rcu(); | |
3038 | ||
3410f22e | 3039 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 3040 | } |
c6d409cf | 3041 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 3042 | |
bf2ae2e4 XL |
3043 | bool sock_is_registered(int family) |
3044 | { | |
66b51b0a | 3045 | return family < NPROTO && rcu_access_pointer(net_families[family]); |
bf2ae2e4 XL |
3046 | } |
3047 | ||
77d76ea3 | 3048 | static int __init sock_init(void) |
1da177e4 | 3049 | { |
b3e19d92 | 3050 | int err; |
2ca794e5 EB |
3051 | /* |
3052 | * Initialize the network sysctl infrastructure. | |
3053 | */ | |
3054 | err = net_sysctl_init(); | |
3055 | if (err) | |
3056 | goto out; | |
b3e19d92 | 3057 | |
1da177e4 | 3058 | /* |
89bddce5 | 3059 | * Initialize skbuff SLAB cache |
1da177e4 LT |
3060 | */ |
3061 | skb_init(); | |
1da177e4 LT |
3062 | |
3063 | /* | |
89bddce5 | 3064 | * Initialize the protocols module. |
1da177e4 LT |
3065 | */ |
3066 | ||
3067 | init_inodecache(); | |
b3e19d92 NP |
3068 | |
3069 | err = register_filesystem(&sock_fs_type); | |
3070 | if (err) | |
3071 | goto out_fs; | |
1da177e4 | 3072 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
3073 | if (IS_ERR(sock_mnt)) { |
3074 | err = PTR_ERR(sock_mnt); | |
3075 | goto out_mount; | |
3076 | } | |
77d76ea3 AK |
3077 | |
3078 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
3079 | */ |
3080 | ||
3081 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
3082 | err = netfilter_init(); |
3083 | if (err) | |
3084 | goto out; | |
1da177e4 | 3085 | #endif |
cbeb321a | 3086 | |
408eccce | 3087 | ptp_classifier_init(); |
c1f19b51 | 3088 | |
b3e19d92 NP |
3089 | out: |
3090 | return err; | |
3091 | ||
3092 | out_mount: | |
3093 | unregister_filesystem(&sock_fs_type); | |
3094 | out_fs: | |
3095 | goto out; | |
1da177e4 LT |
3096 | } |
3097 | ||
77d76ea3 AK |
3098 | core_initcall(sock_init); /* early initcall */ |
3099 | ||
1da177e4 LT |
3100 | #ifdef CONFIG_PROC_FS |
3101 | void socket_seq_show(struct seq_file *seq) | |
3102 | { | |
648845ab TZ |
3103 | seq_printf(seq, "sockets: used %d\n", |
3104 | sock_inuse_get(seq->private)); | |
1da177e4 | 3105 | } |
89bddce5 | 3106 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 3107 | |
89bbfc95 | 3108 | #ifdef CONFIG_COMPAT |
36fd633e | 3109 | static int compat_dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 3110 | { |
6b96018b | 3111 | struct compat_ifconf ifc32; |
7a229387 | 3112 | struct ifconf ifc; |
7a229387 AB |
3113 | int err; |
3114 | ||
6b96018b | 3115 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
3116 | return -EFAULT; |
3117 | ||
36fd633e AV |
3118 | ifc.ifc_len = ifc32.ifc_len; |
3119 | ifc.ifc_req = compat_ptr(ifc32.ifcbuf); | |
7a229387 | 3120 | |
36fd633e AV |
3121 | rtnl_lock(); |
3122 | err = dev_ifconf(net, &ifc, sizeof(struct compat_ifreq)); | |
3123 | rtnl_unlock(); | |
7a229387 AB |
3124 | if (err) |
3125 | return err; | |
3126 | ||
36fd633e | 3127 | ifc32.ifc_len = ifc.ifc_len; |
6b96018b | 3128 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
3129 | return -EFAULT; |
3130 | ||
3131 | return 0; | |
3132 | } | |
3133 | ||
6b96018b | 3134 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 3135 | { |
3a7da39d BH |
3136 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
3137 | bool convert_in = false, convert_out = false; | |
44c02a2c AV |
3138 | size_t buf_size = 0; |
3139 | struct ethtool_rxnfc __user *rxnfc = NULL; | |
3140 | struct ifreq ifr; | |
3a7da39d BH |
3141 | u32 rule_cnt = 0, actual_rule_cnt; |
3142 | u32 ethcmd; | |
7a229387 | 3143 | u32 data; |
3a7da39d | 3144 | int ret; |
7a229387 | 3145 | |
3a7da39d BH |
3146 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
3147 | return -EFAULT; | |
7a229387 | 3148 | |
3a7da39d BH |
3149 | compat_rxnfc = compat_ptr(data); |
3150 | ||
3151 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
3152 | return -EFAULT; |
3153 | ||
3a7da39d BH |
3154 | /* Most ethtool structures are defined without padding. |
3155 | * Unfortunately struct ethtool_rxnfc is an exception. | |
3156 | */ | |
3157 | switch (ethcmd) { | |
3158 | default: | |
3159 | break; | |
3160 | case ETHTOOL_GRXCLSRLALL: | |
3161 | /* Buffer size is variable */ | |
3162 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
3163 | return -EFAULT; | |
3164 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
3165 | return -ENOMEM; | |
3166 | buf_size += rule_cnt * sizeof(u32); | |
3167 | /* fall through */ | |
3168 | case ETHTOOL_GRXRINGS: | |
3169 | case ETHTOOL_GRXCLSRLCNT: | |
3170 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 3171 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
3172 | convert_out = true; |
3173 | /* fall through */ | |
3174 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
3175 | buf_size += sizeof(struct ethtool_rxnfc); |
3176 | convert_in = true; | |
44c02a2c | 3177 | rxnfc = compat_alloc_user_space(buf_size); |
3a7da39d BH |
3178 | break; |
3179 | } | |
3180 | ||
44c02a2c | 3181 | if (copy_from_user(&ifr.ifr_name, &ifr32->ifr_name, IFNAMSIZ)) |
7a229387 AB |
3182 | return -EFAULT; |
3183 | ||
44c02a2c | 3184 | ifr.ifr_data = convert_in ? rxnfc : (void __user *)compat_rxnfc; |
7a229387 | 3185 | |
3a7da39d | 3186 | if (convert_in) { |
127fe533 | 3187 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
3188 | * fs.ring_cookie and at the end of fs, but nowhere else. |
3189 | */ | |
127fe533 AD |
3190 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
3191 | sizeof(compat_rxnfc->fs.m_ext) != | |
3192 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
3193 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
3194 | BUILD_BUG_ON( |
3195 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
3196 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
3197 | offsetof(struct ethtool_rxnfc, fs.location) - | |
3198 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
3199 | ||
3200 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
3201 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
3202 | (void __user *)rxnfc) || | |
3a7da39d BH |
3203 | copy_in_user(&rxnfc->fs.ring_cookie, |
3204 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 | 3205 | (void __user *)(&rxnfc->fs.location + 1) - |
b6168562 WW |
3206 | (void __user *)&rxnfc->fs.ring_cookie)) |
3207 | return -EFAULT; | |
3208 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3209 | if (put_user(rule_cnt, &rxnfc->rule_cnt)) | |
3210 | return -EFAULT; | |
3211 | } else if (copy_in_user(&rxnfc->rule_cnt, | |
3212 | &compat_rxnfc->rule_cnt, | |
3213 | sizeof(rxnfc->rule_cnt))) | |
3a7da39d BH |
3214 | return -EFAULT; |
3215 | } | |
3216 | ||
44c02a2c | 3217 | ret = dev_ioctl(net, SIOCETHTOOL, &ifr, NULL); |
3a7da39d BH |
3218 | if (ret) |
3219 | return ret; | |
3220 | ||
3221 | if (convert_out) { | |
3222 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
3223 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
3224 | (const void __user *)rxnfc) || | |
3a7da39d BH |
3225 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
3226 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
3227 | (const void __user *)(&rxnfc->fs.location + 1) - |
3228 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
3229 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
3230 | sizeof(rxnfc->rule_cnt))) | |
3231 | return -EFAULT; | |
3232 | ||
3233 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3234 | /* As an optimisation, we only copy the actual | |
3235 | * number of rules that the underlying | |
3236 | * function returned. Since Mallory might | |
3237 | * change the rule count in user memory, we | |
3238 | * check that it is less than the rule count | |
3239 | * originally given (as the user buffer size), | |
3240 | * which has been range-checked. | |
3241 | */ | |
3242 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
3243 | return -EFAULT; | |
3244 | if (actual_rule_cnt < rule_cnt) | |
3245 | rule_cnt = actual_rule_cnt; | |
3246 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
3247 | &rxnfc->rule_locs[0], | |
3248 | rule_cnt * sizeof(u32))) | |
3249 | return -EFAULT; | |
3250 | } | |
3251 | } | |
3252 | ||
3253 | return 0; | |
7a229387 AB |
3254 | } |
3255 | ||
7a50a240 AB |
3256 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
3257 | { | |
7a50a240 | 3258 | compat_uptr_t uptr32; |
44c02a2c AV |
3259 | struct ifreq ifr; |
3260 | void __user *saved; | |
3261 | int err; | |
7a50a240 | 3262 | |
44c02a2c | 3263 | if (copy_from_user(&ifr, uifr32, sizeof(struct compat_ifreq))) |
7a50a240 AB |
3264 | return -EFAULT; |
3265 | ||
3266 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
3267 | return -EFAULT; | |
3268 | ||
44c02a2c AV |
3269 | saved = ifr.ifr_settings.ifs_ifsu.raw_hdlc; |
3270 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = compat_ptr(uptr32); | |
7a229387 | 3271 | |
44c02a2c AV |
3272 | err = dev_ioctl(net, SIOCWANDEV, &ifr, NULL); |
3273 | if (!err) { | |
3274 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = saved; | |
3275 | if (copy_to_user(uifr32, &ifr, sizeof(struct compat_ifreq))) | |
3276 | err = -EFAULT; | |
ccbd6a5a | 3277 | } |
44c02a2c | 3278 | return err; |
7a229387 AB |
3279 | } |
3280 | ||
590d4693 BH |
3281 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
3282 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 3283 | struct compat_ifreq __user *u_ifreq32) |
7a229387 | 3284 | { |
44c02a2c | 3285 | struct ifreq ifreq; |
7a229387 AB |
3286 | u32 data32; |
3287 | ||
44c02a2c | 3288 | if (copy_from_user(ifreq.ifr_name, u_ifreq32->ifr_name, IFNAMSIZ)) |
7a229387 | 3289 | return -EFAULT; |
44c02a2c | 3290 | if (get_user(data32, &u_ifreq32->ifr_data)) |
7a229387 | 3291 | return -EFAULT; |
44c02a2c | 3292 | ifreq.ifr_data = compat_ptr(data32); |
7a229387 | 3293 | |
44c02a2c | 3294 | return dev_ioctl(net, cmd, &ifreq, NULL); |
7a229387 AB |
3295 | } |
3296 | ||
37ac39bd JB |
3297 | static int compat_ifreq_ioctl(struct net *net, struct socket *sock, |
3298 | unsigned int cmd, | |
3299 | struct compat_ifreq __user *uifr32) | |
3300 | { | |
3301 | struct ifreq __user *uifr; | |
3302 | int err; | |
3303 | ||
3304 | /* Handle the fact that while struct ifreq has the same *layout* on | |
3305 | * 32/64 for everything but ifreq::ifru_ifmap and ifreq::ifru_data, | |
3306 | * which are handled elsewhere, it still has different *size* due to | |
3307 | * ifreq::ifru_ifmap (which is 16 bytes on 32 bit, 24 bytes on 64-bit, | |
3308 | * resulting in struct ifreq being 32 and 40 bytes respectively). | |
3309 | * As a result, if the struct happens to be at the end of a page and | |
3310 | * the next page isn't readable/writable, we get a fault. To prevent | |
3311 | * that, copy back and forth to the full size. | |
3312 | */ | |
3313 | ||
3314 | uifr = compat_alloc_user_space(sizeof(*uifr)); | |
3315 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
3316 | return -EFAULT; | |
3317 | ||
3318 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
3319 | ||
3320 | if (!err) { | |
3321 | switch (cmd) { | |
3322 | case SIOCGIFFLAGS: | |
3323 | case SIOCGIFMETRIC: | |
3324 | case SIOCGIFMTU: | |
3325 | case SIOCGIFMEM: | |
3326 | case SIOCGIFHWADDR: | |
3327 | case SIOCGIFINDEX: | |
3328 | case SIOCGIFADDR: | |
3329 | case SIOCGIFBRDADDR: | |
3330 | case SIOCGIFDSTADDR: | |
3331 | case SIOCGIFNETMASK: | |
3332 | case SIOCGIFPFLAGS: | |
3333 | case SIOCGIFTXQLEN: | |
3334 | case SIOCGMIIPHY: | |
3335 | case SIOCGMIIREG: | |
c6c9fee3 | 3336 | case SIOCGIFNAME: |
37ac39bd JB |
3337 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
3338 | err = -EFAULT; | |
3339 | break; | |
3340 | } | |
3341 | } | |
3342 | return err; | |
3343 | } | |
3344 | ||
a2116ed2 AB |
3345 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
3346 | struct compat_ifreq __user *uifr32) | |
3347 | { | |
3348 | struct ifreq ifr; | |
3349 | struct compat_ifmap __user *uifmap32; | |
a2116ed2 AB |
3350 | int err; |
3351 | ||
3352 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
3353 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3354 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3355 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3356 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3357 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3358 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3359 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3360 | if (err) |
3361 | return -EFAULT; | |
3362 | ||
44c02a2c | 3363 | err = dev_ioctl(net, cmd, &ifr, NULL); |
a2116ed2 AB |
3364 | |
3365 | if (cmd == SIOCGIFMAP && !err) { | |
3366 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3367 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3368 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3369 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3370 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3371 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3372 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3373 | if (err) |
3374 | err = -EFAULT; | |
3375 | } | |
3376 | return err; | |
3377 | } | |
3378 | ||
7a229387 | 3379 | struct rtentry32 { |
c6d409cf | 3380 | u32 rt_pad1; |
7a229387 AB |
3381 | struct sockaddr rt_dst; /* target address */ |
3382 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
3383 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
3384 | unsigned short rt_flags; |
3385 | short rt_pad2; | |
3386 | u32 rt_pad3; | |
3387 | unsigned char rt_tos; | |
3388 | unsigned char rt_class; | |
3389 | short rt_pad4; | |
3390 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 3391 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
3392 | u32 rt_mtu; /* per route MTU/Window */ |
3393 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
3394 | unsigned short rt_irtt; /* Initial RTT */ |
3395 | }; | |
3396 | ||
3397 | struct in6_rtmsg32 { | |
3398 | struct in6_addr rtmsg_dst; | |
3399 | struct in6_addr rtmsg_src; | |
3400 | struct in6_addr rtmsg_gateway; | |
3401 | u32 rtmsg_type; | |
3402 | u16 rtmsg_dst_len; | |
3403 | u16 rtmsg_src_len; | |
3404 | u32 rtmsg_metric; | |
3405 | u32 rtmsg_info; | |
3406 | u32 rtmsg_flags; | |
3407 | s32 rtmsg_ifindex; | |
3408 | }; | |
3409 | ||
6b96018b AB |
3410 | static int routing_ioctl(struct net *net, struct socket *sock, |
3411 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3412 | { |
3413 | int ret; | |
3414 | void *r = NULL; | |
3415 | struct in6_rtmsg r6; | |
3416 | struct rtentry r4; | |
3417 | char devname[16]; | |
3418 | u32 rtdev; | |
3419 | mm_segment_t old_fs = get_fs(); | |
3420 | ||
6b96018b AB |
3421 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3422 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3423 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3424 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
3425 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3426 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3427 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3428 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3429 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3430 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3431 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3432 | |
3433 | r = (void *) &r6; | |
3434 | } else { /* ipv4 */ | |
6b96018b | 3435 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3436 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3437 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
3438 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
3439 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3440 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3441 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
3442 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3443 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3444 | if (rtdev) { |
c6d409cf | 3445 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3446 | r4.rt_dev = (char __user __force *)devname; |
3447 | devname[15] = 0; | |
7a229387 AB |
3448 | } else |
3449 | r4.rt_dev = NULL; | |
3450 | ||
3451 | r = (void *) &r4; | |
3452 | } | |
3453 | ||
3454 | if (ret) { | |
3455 | ret = -EFAULT; | |
3456 | goto out; | |
3457 | } | |
3458 | ||
c6d409cf | 3459 | set_fs(KERNEL_DS); |
63ff03ab | 3460 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3461 | set_fs(old_fs); |
7a229387 AB |
3462 | |
3463 | out: | |
7a229387 AB |
3464 | return ret; |
3465 | } | |
3466 | ||
3467 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3468 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3469 | * use compatible ioctls |
7a229387 | 3470 | */ |
6b96018b | 3471 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3472 | { |
6b96018b | 3473 | compat_ulong_t tmp; |
7a229387 | 3474 | |
6b96018b | 3475 | if (get_user(tmp, argp)) |
7a229387 AB |
3476 | return -EFAULT; |
3477 | if (tmp == BRCTL_GET_VERSION) | |
3478 | return BRCTL_VERSION + 1; | |
3479 | return -EINVAL; | |
3480 | } | |
3481 | ||
6b96018b AB |
3482 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3483 | unsigned int cmd, unsigned long arg) | |
3484 | { | |
3485 | void __user *argp = compat_ptr(arg); | |
3486 | struct sock *sk = sock->sk; | |
3487 | struct net *net = sock_net(sk); | |
7a229387 | 3488 | |
6b96018b | 3489 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3490 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3491 | |
3492 | switch (cmd) { | |
3493 | case SIOCSIFBR: | |
3494 | case SIOCGIFBR: | |
3495 | return old_bridge_ioctl(argp); | |
6b96018b | 3496 | case SIOCGIFCONF: |
36fd633e | 3497 | return compat_dev_ifconf(net, argp); |
6b96018b AB |
3498 | case SIOCETHTOOL: |
3499 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3500 | case SIOCWANDEV: |
3501 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3502 | case SIOCGIFMAP: |
3503 | case SIOCSIFMAP: | |
3504 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3505 | case SIOCADDRT: |
3506 | case SIOCDELRT: | |
3507 | return routing_ioctl(net, sock, cmd, argp); | |
0768e170 AB |
3508 | case SIOCGSTAMP_OLD: |
3509 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
3510 | if (!sock->ops->gettstamp) |
3511 | return -ENOIOCTLCMD; | |
0768e170 | 3512 | return sock->ops->gettstamp(sock, argp, cmd == SIOCGSTAMP_OLD, |
c7cbdbf2 AB |
3513 | !COMPAT_USE_64BIT_TIME); |
3514 | ||
590d4693 BH |
3515 | case SIOCBONDSLAVEINFOQUERY: |
3516 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3517 | case SIOCSHWTSTAMP: |
fd468c74 | 3518 | case SIOCGHWTSTAMP: |
590d4693 | 3519 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3520 | |
3521 | case FIOSETOWN: | |
3522 | case SIOCSPGRP: | |
3523 | case FIOGETOWN: | |
3524 | case SIOCGPGRP: | |
3525 | case SIOCBRADDBR: | |
3526 | case SIOCBRDELBR: | |
3527 | case SIOCGIFVLAN: | |
3528 | case SIOCSIFVLAN: | |
3529 | case SIOCADDDLCI: | |
3530 | case SIOCDELDLCI: | |
c62cce2c | 3531 | case SIOCGSKNS: |
0768e170 AB |
3532 | case SIOCGSTAMP_NEW: |
3533 | case SIOCGSTAMPNS_NEW: | |
6b96018b AB |
3534 | return sock_ioctl(file, cmd, arg); |
3535 | ||
3536 | case SIOCGIFFLAGS: | |
3537 | case SIOCSIFFLAGS: | |
3538 | case SIOCGIFMETRIC: | |
3539 | case SIOCSIFMETRIC: | |
3540 | case SIOCGIFMTU: | |
3541 | case SIOCSIFMTU: | |
3542 | case SIOCGIFMEM: | |
3543 | case SIOCSIFMEM: | |
3544 | case SIOCGIFHWADDR: | |
3545 | case SIOCSIFHWADDR: | |
3546 | case SIOCADDMULTI: | |
3547 | case SIOCDELMULTI: | |
3548 | case SIOCGIFINDEX: | |
6b96018b AB |
3549 | case SIOCGIFADDR: |
3550 | case SIOCSIFADDR: | |
3551 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3552 | case SIOCDIFADDR: |
6b96018b AB |
3553 | case SIOCGIFBRDADDR: |
3554 | case SIOCSIFBRDADDR: | |
3555 | case SIOCGIFDSTADDR: | |
3556 | case SIOCSIFDSTADDR: | |
3557 | case SIOCGIFNETMASK: | |
3558 | case SIOCSIFNETMASK: | |
3559 | case SIOCSIFPFLAGS: | |
3560 | case SIOCGIFPFLAGS: | |
3561 | case SIOCGIFTXQLEN: | |
3562 | case SIOCSIFTXQLEN: | |
3563 | case SIOCBRADDIF: | |
3564 | case SIOCBRDELIF: | |
c6c9fee3 | 3565 | case SIOCGIFNAME: |
9177efd3 AB |
3566 | case SIOCSIFNAME: |
3567 | case SIOCGMIIPHY: | |
3568 | case SIOCGMIIREG: | |
3569 | case SIOCSMIIREG: | |
f92d4fc9 AV |
3570 | case SIOCBONDENSLAVE: |
3571 | case SIOCBONDRELEASE: | |
3572 | case SIOCBONDSETHWADDR: | |
3573 | case SIOCBONDCHANGEACTIVE: | |
37ac39bd JB |
3574 | return compat_ifreq_ioctl(net, sock, cmd, argp); |
3575 | ||
6b96018b AB |
3576 | case SIOCSARP: |
3577 | case SIOCGARP: | |
3578 | case SIOCDARP: | |
c7dc504e | 3579 | case SIOCOUTQ: |
9d7bf41f | 3580 | case SIOCOUTQNSD: |
6b96018b | 3581 | case SIOCATMARK: |
63ff03ab | 3582 | return sock_do_ioctl(net, sock, cmd, arg); |
9177efd3 AB |
3583 | } |
3584 | ||
6b96018b AB |
3585 | return -ENOIOCTLCMD; |
3586 | } | |
7a229387 | 3587 | |
95c96174 | 3588 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3589 | unsigned long arg) |
89bbfc95 SP |
3590 | { |
3591 | struct socket *sock = file->private_data; | |
3592 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3593 | struct sock *sk; |
3594 | struct net *net; | |
3595 | ||
3596 | sk = sock->sk; | |
3597 | net = sock_net(sk); | |
89bbfc95 SP |
3598 | |
3599 | if (sock->ops->compat_ioctl) | |
3600 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3601 | ||
87de87d5 DM |
3602 | if (ret == -ENOIOCTLCMD && |
3603 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3604 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3605 | ||
6b96018b AB |
3606 | if (ret == -ENOIOCTLCMD) |
3607 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3608 | ||
89bbfc95 SP |
3609 | return ret; |
3610 | } | |
3611 | #endif | |
3612 | ||
8a3c245c PT |
3613 | /** |
3614 | * kernel_bind - bind an address to a socket (kernel space) | |
3615 | * @sock: socket | |
3616 | * @addr: address | |
3617 | * @addrlen: length of address | |
3618 | * | |
3619 | * Returns 0 or an error. | |
3620 | */ | |
3621 | ||
ac5a488e SS |
3622 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3623 | { | |
3624 | return sock->ops->bind(sock, addr, addrlen); | |
3625 | } | |
c6d409cf | 3626 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e | 3627 | |
8a3c245c PT |
3628 | /** |
3629 | * kernel_listen - move socket to listening state (kernel space) | |
3630 | * @sock: socket | |
3631 | * @backlog: pending connections queue size | |
3632 | * | |
3633 | * Returns 0 or an error. | |
3634 | */ | |
3635 | ||
ac5a488e SS |
3636 | int kernel_listen(struct socket *sock, int backlog) |
3637 | { | |
3638 | return sock->ops->listen(sock, backlog); | |
3639 | } | |
c6d409cf | 3640 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e | 3641 | |
8a3c245c PT |
3642 | /** |
3643 | * kernel_accept - accept a connection (kernel space) | |
3644 | * @sock: listening socket | |
3645 | * @newsock: new connected socket | |
3646 | * @flags: flags | |
3647 | * | |
3648 | * @flags must be SOCK_CLOEXEC, SOCK_NONBLOCK or 0. | |
3649 | * If it fails, @newsock is guaranteed to be %NULL. | |
3650 | * Returns 0 or an error. | |
3651 | */ | |
3652 | ||
ac5a488e SS |
3653 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) |
3654 | { | |
3655 | struct sock *sk = sock->sk; | |
3656 | int err; | |
3657 | ||
3658 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3659 | newsock); | |
3660 | if (err < 0) | |
3661 | goto done; | |
3662 | ||
cdfbabfb | 3663 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3664 | if (err < 0) { |
3665 | sock_release(*newsock); | |
fa8705b0 | 3666 | *newsock = NULL; |
ac5a488e SS |
3667 | goto done; |
3668 | } | |
3669 | ||
3670 | (*newsock)->ops = sock->ops; | |
1b08534e | 3671 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3672 | |
3673 | done: | |
3674 | return err; | |
3675 | } | |
c6d409cf | 3676 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e | 3677 | |
8a3c245c PT |
3678 | /** |
3679 | * kernel_connect - connect a socket (kernel space) | |
3680 | * @sock: socket | |
3681 | * @addr: address | |
3682 | * @addrlen: address length | |
3683 | * @flags: flags (O_NONBLOCK, ...) | |
3684 | * | |
3685 | * For datagram sockets, @addr is the addres to which datagrams are sent | |
3686 | * by default, and the only address from which datagrams are received. | |
3687 | * For stream sockets, attempts to connect to @addr. | |
3688 | * Returns 0 or an error code. | |
3689 | */ | |
3690 | ||
ac5a488e | 3691 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, |
4768fbcb | 3692 | int flags) |
ac5a488e SS |
3693 | { |
3694 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3695 | } | |
c6d409cf | 3696 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e | 3697 | |
8a3c245c PT |
3698 | /** |
3699 | * kernel_getsockname - get the address which the socket is bound (kernel space) | |
3700 | * @sock: socket | |
3701 | * @addr: address holder | |
3702 | * | |
3703 | * Fills the @addr pointer with the address which the socket is bound. | |
3704 | * Returns 0 or an error code. | |
3705 | */ | |
3706 | ||
9b2c45d4 | 3707 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3708 | { |
9b2c45d4 | 3709 | return sock->ops->getname(sock, addr, 0); |
ac5a488e | 3710 | } |
c6d409cf | 3711 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e | 3712 | |
8a3c245c PT |
3713 | /** |
3714 | * kernel_peername - get the address which the socket is connected (kernel space) | |
3715 | * @sock: socket | |
3716 | * @addr: address holder | |
3717 | * | |
3718 | * Fills the @addr pointer with the address which the socket is connected. | |
3719 | * Returns 0 or an error code. | |
3720 | */ | |
3721 | ||
9b2c45d4 | 3722 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3723 | { |
9b2c45d4 | 3724 | return sock->ops->getname(sock, addr, 1); |
ac5a488e | 3725 | } |
c6d409cf | 3726 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e | 3727 | |
8a3c245c PT |
3728 | /** |
3729 | * kernel_getsockopt - get a socket option (kernel space) | |
3730 | * @sock: socket | |
3731 | * @level: API level (SOL_SOCKET, ...) | |
3732 | * @optname: option tag | |
3733 | * @optval: option value | |
3734 | * @optlen: option length | |
3735 | * | |
3736 | * Assigns the option length to @optlen. | |
3737 | * Returns 0 or an error. | |
3738 | */ | |
3739 | ||
ac5a488e SS |
3740 | int kernel_getsockopt(struct socket *sock, int level, int optname, |
3741 | char *optval, int *optlen) | |
3742 | { | |
3743 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3744 | char __user *uoptval; |
3745 | int __user *uoptlen; | |
ac5a488e SS |
3746 | int err; |
3747 | ||
fb8621bb NK |
3748 | uoptval = (char __user __force *) optval; |
3749 | uoptlen = (int __user __force *) optlen; | |
3750 | ||
ac5a488e SS |
3751 | set_fs(KERNEL_DS); |
3752 | if (level == SOL_SOCKET) | |
fb8621bb | 3753 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3754 | else |
fb8621bb NK |
3755 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3756 | uoptlen); | |
ac5a488e SS |
3757 | set_fs(oldfs); |
3758 | return err; | |
3759 | } | |
c6d409cf | 3760 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e | 3761 | |
8a3c245c PT |
3762 | /** |
3763 | * kernel_setsockopt - set a socket option (kernel space) | |
3764 | * @sock: socket | |
3765 | * @level: API level (SOL_SOCKET, ...) | |
3766 | * @optname: option tag | |
3767 | * @optval: option value | |
3768 | * @optlen: option length | |
3769 | * | |
3770 | * Returns 0 or an error. | |
3771 | */ | |
3772 | ||
ac5a488e | 3773 | int kernel_setsockopt(struct socket *sock, int level, int optname, |
b7058842 | 3774 | char *optval, unsigned int optlen) |
ac5a488e SS |
3775 | { |
3776 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3777 | char __user *uoptval; |
ac5a488e SS |
3778 | int err; |
3779 | ||
fb8621bb NK |
3780 | uoptval = (char __user __force *) optval; |
3781 | ||
ac5a488e SS |
3782 | set_fs(KERNEL_DS); |
3783 | if (level == SOL_SOCKET) | |
fb8621bb | 3784 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3785 | else |
fb8621bb | 3786 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3787 | optlen); |
3788 | set_fs(oldfs); | |
3789 | return err; | |
3790 | } | |
c6d409cf | 3791 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e | 3792 | |
8a3c245c PT |
3793 | /** |
3794 | * kernel_sendpage - send a &page through a socket (kernel space) | |
3795 | * @sock: socket | |
3796 | * @page: page | |
3797 | * @offset: page offset | |
3798 | * @size: total size in bytes | |
3799 | * @flags: flags (MSG_DONTWAIT, ...) | |
3800 | * | |
3801 | * Returns the total amount sent in bytes or an error. | |
3802 | */ | |
3803 | ||
ac5a488e SS |
3804 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, |
3805 | size_t size, int flags) | |
3806 | { | |
3807 | if (sock->ops->sendpage) | |
3808 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3809 | ||
3810 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3811 | } | |
c6d409cf | 3812 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3813 | |
8a3c245c PT |
3814 | /** |
3815 | * kernel_sendpage_locked - send a &page through the locked sock (kernel space) | |
3816 | * @sk: sock | |
3817 | * @page: page | |
3818 | * @offset: page offset | |
3819 | * @size: total size in bytes | |
3820 | * @flags: flags (MSG_DONTWAIT, ...) | |
3821 | * | |
3822 | * Returns the total amount sent in bytes or an error. | |
3823 | * Caller must hold @sk. | |
3824 | */ | |
3825 | ||
306b13eb TH |
3826 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3827 | size_t size, int flags) | |
3828 | { | |
3829 | struct socket *sock = sk->sk_socket; | |
3830 | ||
3831 | if (sock->ops->sendpage_locked) | |
3832 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3833 | flags); | |
3834 | ||
3835 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3836 | } | |
3837 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3838 | ||
8a3c245c PT |
3839 | /** |
3840 | * kernel_shutdown - shut down part of a full-duplex connection (kernel space) | |
3841 | * @sock: socket | |
3842 | * @how: connection part | |
3843 | * | |
3844 | * Returns 0 or an error. | |
3845 | */ | |
3846 | ||
91cf45f0 TM |
3847 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3848 | { | |
3849 | return sock->ops->shutdown(sock, how); | |
3850 | } | |
91cf45f0 | 3851 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 | 3852 | |
8a3c245c PT |
3853 | /** |
3854 | * kernel_sock_ip_overhead - returns the IP overhead imposed by a socket | |
3855 | * @sk: socket | |
3856 | * | |
3857 | * This routine returns the IP overhead imposed by a socket i.e. | |
3858 | * the length of the underlying IP header, depending on whether | |
3859 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
3860 | * on at the socket. Assumes that the caller has a lock on the socket. | |
113c3075 | 3861 | */ |
8a3c245c | 3862 | |
113c3075 P |
3863 | u32 kernel_sock_ip_overhead(struct sock *sk) |
3864 | { | |
3865 | struct inet_sock *inet; | |
3866 | struct ip_options_rcu *opt; | |
3867 | u32 overhead = 0; | |
113c3075 P |
3868 | #if IS_ENABLED(CONFIG_IPV6) |
3869 | struct ipv6_pinfo *np; | |
3870 | struct ipv6_txoptions *optv6 = NULL; | |
3871 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3872 | ||
3873 | if (!sk) | |
3874 | return overhead; | |
3875 | ||
113c3075 P |
3876 | switch (sk->sk_family) { |
3877 | case AF_INET: | |
3878 | inet = inet_sk(sk); | |
3879 | overhead += sizeof(struct iphdr); | |
3880 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3881 | sock_owned_by_user(sk)); |
113c3075 P |
3882 | if (opt) |
3883 | overhead += opt->opt.optlen; | |
3884 | return overhead; | |
3885 | #if IS_ENABLED(CONFIG_IPV6) | |
3886 | case AF_INET6: | |
3887 | np = inet6_sk(sk); | |
3888 | overhead += sizeof(struct ipv6hdr); | |
3889 | if (np) | |
3890 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3891 | sock_owned_by_user(sk)); |
113c3075 P |
3892 | if (optv6) |
3893 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3894 | return overhead; | |
3895 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3896 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3897 | return overhead; | |
3898 | } | |
3899 | } | |
3900 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |