Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * NET An implementation of the SOCKET network access protocol. | |
3 | * | |
4 | * Version: @(#)socket.c 1.1.93 18/02/95 | |
5 | * | |
6 | * Authors: Orest Zborowski, <obz@Kodak.COM> | |
02c30a84 | 7 | * Ross Biro |
1da177e4 LT |
8 | * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> |
9 | * | |
10 | * Fixes: | |
11 | * Anonymous : NOTSOCK/BADF cleanup. Error fix in | |
12 | * shutdown() | |
13 | * Alan Cox : verify_area() fixes | |
14 | * Alan Cox : Removed DDI | |
15 | * Jonathan Kamens : SOCK_DGRAM reconnect bug | |
16 | * Alan Cox : Moved a load of checks to the very | |
17 | * top level. | |
18 | * Alan Cox : Move address structures to/from user | |
19 | * mode above the protocol layers. | |
20 | * Rob Janssen : Allow 0 length sends. | |
21 | * Alan Cox : Asynchronous I/O support (cribbed from the | |
22 | * tty drivers). | |
23 | * Niibe Yutaka : Asynchronous I/O for writes (4.4BSD style) | |
24 | * Jeff Uphoff : Made max number of sockets command-line | |
25 | * configurable. | |
26 | * Matti Aarnio : Made the number of sockets dynamic, | |
27 | * to be allocated when needed, and mr. | |
28 | * Uphoff's max is used as max to be | |
29 | * allowed to allocate. | |
30 | * Linus : Argh. removed all the socket allocation | |
31 | * altogether: it's in the inode now. | |
32 | * Alan Cox : Made sock_alloc()/sock_release() public | |
33 | * for NetROM and future kernel nfsd type | |
34 | * stuff. | |
35 | * Alan Cox : sendmsg/recvmsg basics. | |
36 | * Tom Dyas : Export net symbols. | |
37 | * Marcin Dalecki : Fixed problems with CONFIG_NET="n". | |
38 | * Alan Cox : Added thread locking to sys_* calls | |
39 | * for sockets. May have errors at the | |
40 | * moment. | |
41 | * Kevin Buhr : Fixed the dumb errors in the above. | |
42 | * Andi Kleen : Some small cleanups, optimizations, | |
43 | * and fixed a copy_from_user() bug. | |
44 | * Tigran Aivazian : sys_send(args) calls sys_sendto(args, NULL, 0) | |
89bddce5 | 45 | * Tigran Aivazian : Made listen(2) backlog sanity checks |
1da177e4 LT |
46 | * protocol-independent |
47 | * | |
48 | * | |
49 | * This program is free software; you can redistribute it and/or | |
50 | * modify it under the terms of the GNU General Public License | |
51 | * as published by the Free Software Foundation; either version | |
52 | * 2 of the License, or (at your option) any later version. | |
53 | * | |
54 | * | |
55 | * This module is effectively the top level interface to the BSD socket | |
89bddce5 | 56 | * paradigm. |
1da177e4 LT |
57 | * |
58 | * Based upon Swansea University Computer Society NET3.039 | |
59 | */ | |
60 | ||
1da177e4 | 61 | #include <linux/mm.h> |
1da177e4 LT |
62 | #include <linux/socket.h> |
63 | #include <linux/file.h> | |
64 | #include <linux/net.h> | |
65 | #include <linux/interrupt.h> | |
aaca0bdc | 66 | #include <linux/thread_info.h> |
55737fda | 67 | #include <linux/rcupdate.h> |
1da177e4 LT |
68 | #include <linux/netdevice.h> |
69 | #include <linux/proc_fs.h> | |
70 | #include <linux/seq_file.h> | |
4a3e2f71 | 71 | #include <linux/mutex.h> |
1da177e4 | 72 | #include <linux/if_bridge.h> |
20380731 ACM |
73 | #include <linux/if_frad.h> |
74 | #include <linux/if_vlan.h> | |
408eccce | 75 | #include <linux/ptp_classify.h> |
1da177e4 LT |
76 | #include <linux/init.h> |
77 | #include <linux/poll.h> | |
78 | #include <linux/cache.h> | |
79 | #include <linux/module.h> | |
80 | #include <linux/highmem.h> | |
1da177e4 LT |
81 | #include <linux/mount.h> |
82 | #include <linux/security.h> | |
83 | #include <linux/syscalls.h> | |
84 | #include <linux/compat.h> | |
85 | #include <linux/kmod.h> | |
3ec3b2fb | 86 | #include <linux/audit.h> |
d86b5e0e | 87 | #include <linux/wireless.h> |
1b8d7ae4 | 88 | #include <linux/nsproxy.h> |
1fd7317d | 89 | #include <linux/magic.h> |
5a0e3ad6 | 90 | #include <linux/slab.h> |
600e1779 | 91 | #include <linux/xattr.h> |
1da177e4 | 92 | |
7c0f6ba6 | 93 | #include <linux/uaccess.h> |
1da177e4 LT |
94 | #include <asm/unistd.h> |
95 | ||
96 | #include <net/compat.h> | |
87de87d5 | 97 | #include <net/wext.h> |
f8451725 | 98 | #include <net/cls_cgroup.h> |
1da177e4 LT |
99 | |
100 | #include <net/sock.h> | |
101 | #include <linux/netfilter.h> | |
102 | ||
6b96018b AB |
103 | #include <linux/if_tun.h> |
104 | #include <linux/ipv6_route.h> | |
105 | #include <linux/route.h> | |
6b96018b | 106 | #include <linux/sockios.h> |
076bb0c8 | 107 | #include <net/busy_poll.h> |
f24b9be5 | 108 | #include <linux/errqueue.h> |
06021292 | 109 | |
e0d1095a | 110 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
111 | unsigned int sysctl_net_busy_read __read_mostly; |
112 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 113 | #endif |
6b96018b | 114 | |
8ae5e030 AV |
115 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
116 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 117 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
118 | |
119 | static int sock_close(struct inode *inode, struct file *file); | |
15252423 CH |
120 | static struct wait_queue_head *sock_get_poll_head(struct file *file, |
121 | __poll_t events); | |
122 | static __poll_t sock_poll_mask(struct file *file, __poll_t); | |
123 | static __poll_t sock_poll(struct file *file, struct poll_table_struct *wait); | |
89bddce5 | 124 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
125 | #ifdef CONFIG_COMPAT |
126 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 127 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 128 | #endif |
1da177e4 | 129 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
130 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
131 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 132 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 133 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 134 | unsigned int flags); |
1da177e4 | 135 | |
1da177e4 LT |
136 | /* |
137 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
138 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
139 | */ | |
140 | ||
da7071d7 | 141 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
142 | .owner = THIS_MODULE, |
143 | .llseek = no_llseek, | |
8ae5e030 AV |
144 | .read_iter = sock_read_iter, |
145 | .write_iter = sock_write_iter, | |
15252423 CH |
146 | .get_poll_head = sock_get_poll_head, |
147 | .poll_mask = sock_poll_mask, | |
1da177e4 LT |
148 | .poll = sock_poll, |
149 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
150 | #ifdef CONFIG_COMPAT |
151 | .compat_ioctl = compat_sock_ioctl, | |
152 | #endif | |
1da177e4 | 153 | .mmap = sock_mmap, |
1da177e4 LT |
154 | .release = sock_close, |
155 | .fasync = sock_fasync, | |
5274f052 JA |
156 | .sendpage = sock_sendpage, |
157 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 158 | .splice_read = sock_splice_read, |
1da177e4 LT |
159 | }; |
160 | ||
161 | /* | |
162 | * The protocol list. Each protocol is registered in here. | |
163 | */ | |
164 | ||
1da177e4 | 165 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 166 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 167 | |
1da177e4 | 168 | /* |
89bddce5 SH |
169 | * Support routines. |
170 | * Move socket addresses back and forth across the kernel/user | |
171 | * divide and look after the messy bits. | |
1da177e4 LT |
172 | */ |
173 | ||
1da177e4 LT |
174 | /** |
175 | * move_addr_to_kernel - copy a socket address into kernel space | |
176 | * @uaddr: Address in user space | |
177 | * @kaddr: Address in kernel space | |
178 | * @ulen: Length in user space | |
179 | * | |
180 | * The address is copied into kernel space. If the provided address is | |
181 | * too long an error code of -EINVAL is returned. If the copy gives | |
182 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
183 | */ | |
184 | ||
43db362d | 185 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 186 | { |
230b1839 | 187 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 188 | return -EINVAL; |
89bddce5 | 189 | if (ulen == 0) |
1da177e4 | 190 | return 0; |
89bddce5 | 191 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 192 | return -EFAULT; |
3ec3b2fb | 193 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
194 | } |
195 | ||
196 | /** | |
197 | * move_addr_to_user - copy an address to user space | |
198 | * @kaddr: kernel space address | |
199 | * @klen: length of address in kernel | |
200 | * @uaddr: user space address | |
201 | * @ulen: pointer to user length field | |
202 | * | |
203 | * The value pointed to by ulen on entry is the buffer length available. | |
204 | * This is overwritten with the buffer space used. -EINVAL is returned | |
205 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
206 | * is returned if either the buffer or the length field are not | |
207 | * accessible. | |
208 | * After copying the data up to the limit the user specifies, the true | |
209 | * length of the data is written over the length limit the user | |
210 | * specified. Zero is returned for a success. | |
211 | */ | |
89bddce5 | 212 | |
43db362d | 213 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 214 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
215 | { |
216 | int err; | |
217 | int len; | |
218 | ||
68c6beb3 | 219 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
220 | err = get_user(len, ulen); |
221 | if (err) | |
1da177e4 | 222 | return err; |
89bddce5 SH |
223 | if (len > klen) |
224 | len = klen; | |
68c6beb3 | 225 | if (len < 0) |
1da177e4 | 226 | return -EINVAL; |
89bddce5 | 227 | if (len) { |
d6fe3945 SG |
228 | if (audit_sockaddr(klen, kaddr)) |
229 | return -ENOMEM; | |
89bddce5 | 230 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
231 | return -EFAULT; |
232 | } | |
233 | /* | |
89bddce5 SH |
234 | * "fromlen shall refer to the value before truncation.." |
235 | * 1003.1g | |
1da177e4 LT |
236 | */ |
237 | return __put_user(klen, ulen); | |
238 | } | |
239 | ||
08009a76 | 240 | static struct kmem_cache *sock_inode_cachep __ro_after_init; |
1da177e4 LT |
241 | |
242 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
243 | { | |
244 | struct socket_alloc *ei; | |
eaefd110 | 245 | struct socket_wq *wq; |
89bddce5 | 246 | |
e94b1766 | 247 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
248 | if (!ei) |
249 | return NULL; | |
eaefd110 ED |
250 | wq = kmalloc(sizeof(*wq), GFP_KERNEL); |
251 | if (!wq) { | |
43815482 ED |
252 | kmem_cache_free(sock_inode_cachep, ei); |
253 | return NULL; | |
254 | } | |
eaefd110 ED |
255 | init_waitqueue_head(&wq->wait); |
256 | wq->fasync_list = NULL; | |
574aab1e | 257 | wq->flags = 0; |
eaefd110 | 258 | RCU_INIT_POINTER(ei->socket.wq, wq); |
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 | ||
269 | static void sock_destroy_inode(struct inode *inode) | |
270 | { | |
43815482 | 271 | struct socket_alloc *ei; |
eaefd110 | 272 | struct socket_wq *wq; |
43815482 ED |
273 | |
274 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
eaefd110 | 275 | wq = rcu_dereference_protected(ei->socket.wq, 1); |
61845220 | 276 | kfree_rcu(wq, rcu); |
43815482 | 277 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
278 | } |
279 | ||
51cc5068 | 280 | static void init_once(void *foo) |
1da177e4 | 281 | { |
89bddce5 | 282 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 283 | |
a35afb83 | 284 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 285 | } |
89bddce5 | 286 | |
1e911632 | 287 | static void init_inodecache(void) |
1da177e4 LT |
288 | { |
289 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
290 | sizeof(struct socket_alloc), |
291 | 0, | |
292 | (SLAB_HWCACHE_ALIGN | | |
293 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 294 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 295 | init_once); |
1e911632 | 296 | BUG_ON(sock_inode_cachep == NULL); |
1da177e4 LT |
297 | } |
298 | ||
b87221de | 299 | static const struct super_operations sockfs_ops = { |
c6d409cf ED |
300 | .alloc_inode = sock_alloc_inode, |
301 | .destroy_inode = sock_destroy_inode, | |
302 | .statfs = simple_statfs, | |
1da177e4 LT |
303 | }; |
304 | ||
c23fbb6b ED |
305 | /* |
306 | * sockfs_dname() is called from d_path(). | |
307 | */ | |
308 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
309 | { | |
310 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 311 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
312 | } |
313 | ||
3ba13d17 | 314 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 315 | .d_dname = sockfs_dname, |
1da177e4 LT |
316 | }; |
317 | ||
bba0bd31 AG |
318 | static int sockfs_xattr_get(const struct xattr_handler *handler, |
319 | struct dentry *dentry, struct inode *inode, | |
320 | const char *suffix, void *value, size_t size) | |
321 | { | |
322 | if (value) { | |
323 | if (dentry->d_name.len + 1 > size) | |
324 | return -ERANGE; | |
325 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
326 | } | |
327 | return dentry->d_name.len + 1; | |
328 | } | |
329 | ||
330 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" | |
331 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
332 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
333 | ||
334 | static const struct xattr_handler sockfs_xattr_handler = { | |
335 | .name = XATTR_NAME_SOCKPROTONAME, | |
336 | .get = sockfs_xattr_get, | |
337 | }; | |
338 | ||
4a590153 AG |
339 | static int sockfs_security_xattr_set(const struct xattr_handler *handler, |
340 | struct dentry *dentry, struct inode *inode, | |
341 | const char *suffix, const void *value, | |
342 | size_t size, int flags) | |
343 | { | |
344 | /* Handled by LSM. */ | |
345 | return -EAGAIN; | |
346 | } | |
347 | ||
348 | static const struct xattr_handler sockfs_security_xattr_handler = { | |
349 | .prefix = XATTR_SECURITY_PREFIX, | |
350 | .set = sockfs_security_xattr_set, | |
351 | }; | |
352 | ||
bba0bd31 AG |
353 | static const struct xattr_handler *sockfs_xattr_handlers[] = { |
354 | &sockfs_xattr_handler, | |
4a590153 | 355 | &sockfs_security_xattr_handler, |
bba0bd31 AG |
356 | NULL |
357 | }; | |
358 | ||
c74a1cbb AV |
359 | static struct dentry *sockfs_mount(struct file_system_type *fs_type, |
360 | int flags, const char *dev_name, void *data) | |
361 | { | |
bba0bd31 AG |
362 | return mount_pseudo_xattr(fs_type, "socket:", &sockfs_ops, |
363 | sockfs_xattr_handlers, | |
364 | &sockfs_dentry_operations, SOCKFS_MAGIC); | |
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", | |
371 | .mount = sockfs_mount, | |
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 | ||
aab174f0 | 392 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 393 | { |
7cbe66b6 | 394 | struct qstr name = { .name = "" }; |
2c48b9c4 | 395 | struct path path; |
7cbe66b6 | 396 | struct file *file; |
1da177e4 | 397 | |
600e1779 MY |
398 | if (dname) { |
399 | name.name = dname; | |
400 | name.len = strlen(name.name); | |
401 | } else if (sock->sk) { | |
402 | name.name = sock->sk->sk_prot_creator->name; | |
403 | name.len = strlen(name.name); | |
404 | } | |
4b936885 | 405 | path.dentry = d_alloc_pseudo(sock_mnt->mnt_sb, &name); |
8e1611e2 AV |
406 | if (unlikely(!path.dentry)) { |
407 | sock_release(sock); | |
28407630 | 408 | return ERR_PTR(-ENOMEM); |
8e1611e2 | 409 | } |
2c48b9c4 | 410 | path.mnt = mntget(sock_mnt); |
39d8c1b6 | 411 | |
2c48b9c4 | 412 | d_instantiate(path.dentry, SOCK_INODE(sock)); |
39d8c1b6 | 413 | |
2c48b9c4 | 414 | file = alloc_file(&path, FMODE_READ | FMODE_WRITE, |
ce8d2cdf | 415 | &socket_file_ops); |
b5ffe634 | 416 | if (IS_ERR(file)) { |
8e1611e2 | 417 | /* drop dentry, keep inode for a bit */ |
c5ef6035 | 418 | ihold(d_inode(path.dentry)); |
2c48b9c4 | 419 | path_put(&path); |
8e1611e2 AV |
420 | /* ... and now kill it properly */ |
421 | sock_release(sock); | |
39b65252 | 422 | return file; |
cc3808f8 AV |
423 | } |
424 | ||
425 | sock->file = file; | |
77d27200 | 426 | file->f_flags = O_RDWR | (flags & O_NONBLOCK); |
39d8c1b6 | 427 | file->private_data = sock; |
28407630 | 428 | return file; |
39d8c1b6 | 429 | } |
56b31d1c | 430 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 431 | |
56b31d1c | 432 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
433 | { |
434 | struct file *newfile; | |
28407630 | 435 | int fd = get_unused_fd_flags(flags); |
ce4bb04c AV |
436 | if (unlikely(fd < 0)) { |
437 | sock_release(sock); | |
28407630 | 438 | return fd; |
ce4bb04c | 439 | } |
39d8c1b6 | 440 | |
aab174f0 | 441 | newfile = sock_alloc_file(sock, flags, NULL); |
28407630 | 442 | if (likely(!IS_ERR(newfile))) { |
39d8c1b6 | 443 | fd_install(fd, newfile); |
28407630 AV |
444 | return fd; |
445 | } | |
7cbe66b6 | 446 | |
28407630 AV |
447 | put_unused_fd(fd); |
448 | return PTR_ERR(newfile); | |
1da177e4 LT |
449 | } |
450 | ||
406a3c63 | 451 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 452 | { |
6cb153ca BL |
453 | if (file->f_op == &socket_file_ops) |
454 | return file->private_data; /* set in sock_map_fd */ | |
455 | ||
23bb80d2 ED |
456 | *err = -ENOTSOCK; |
457 | return NULL; | |
6cb153ca | 458 | } |
406a3c63 | 459 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 460 | |
1da177e4 | 461 | /** |
c6d409cf | 462 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
463 | * @fd: file handle |
464 | * @err: pointer to an error code return | |
465 | * | |
466 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 467 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
468 | * with a negative errno code and NULL is returned. The function checks |
469 | * for both invalid handles and passing a handle which is not a socket. | |
470 | * | |
471 | * On a success the socket object pointer is returned. | |
472 | */ | |
473 | ||
474 | struct socket *sockfd_lookup(int fd, int *err) | |
475 | { | |
476 | struct file *file; | |
1da177e4 LT |
477 | struct socket *sock; |
478 | ||
89bddce5 SH |
479 | file = fget(fd); |
480 | if (!file) { | |
1da177e4 LT |
481 | *err = -EBADF; |
482 | return NULL; | |
483 | } | |
89bddce5 | 484 | |
6cb153ca BL |
485 | sock = sock_from_file(file, err); |
486 | if (!sock) | |
1da177e4 | 487 | fput(file); |
6cb153ca BL |
488 | return sock; |
489 | } | |
c6d409cf | 490 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 491 | |
6cb153ca BL |
492 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
493 | { | |
00e188ef | 494 | struct fd f = fdget(fd); |
6cb153ca BL |
495 | struct socket *sock; |
496 | ||
3672558c | 497 | *err = -EBADF; |
00e188ef AV |
498 | if (f.file) { |
499 | sock = sock_from_file(f.file, err); | |
500 | if (likely(sock)) { | |
501 | *fput_needed = f.flags; | |
6cb153ca | 502 | return sock; |
00e188ef AV |
503 | } |
504 | fdput(f); | |
1da177e4 | 505 | } |
6cb153ca | 506 | return NULL; |
1da177e4 LT |
507 | } |
508 | ||
600e1779 MY |
509 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
510 | size_t size) | |
511 | { | |
512 | ssize_t len; | |
513 | ssize_t used = 0; | |
514 | ||
c5ef6035 | 515 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
516 | if (len < 0) |
517 | return len; | |
518 | used += len; | |
519 | if (buffer) { | |
520 | if (size < used) | |
521 | return -ERANGE; | |
522 | buffer += len; | |
523 | } | |
524 | ||
525 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
526 | used += len; | |
527 | if (buffer) { | |
528 | if (size < used) | |
529 | return -ERANGE; | |
530 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
531 | buffer += len; | |
532 | } | |
533 | ||
534 | return used; | |
535 | } | |
536 | ||
dc647ec8 | 537 | static int sockfs_setattr(struct dentry *dentry, struct iattr *iattr) |
86741ec2 LC |
538 | { |
539 | int err = simple_setattr(dentry, iattr); | |
540 | ||
e1a3a60a | 541 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
542 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
543 | ||
544 | sock->sk->sk_uid = iattr->ia_uid; | |
545 | } | |
546 | ||
547 | return err; | |
548 | } | |
549 | ||
600e1779 | 550 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 551 | .listxattr = sockfs_listxattr, |
86741ec2 | 552 | .setattr = sockfs_setattr, |
600e1779 MY |
553 | }; |
554 | ||
1da177e4 LT |
555 | /** |
556 | * sock_alloc - allocate a socket | |
89bddce5 | 557 | * |
1da177e4 LT |
558 | * Allocate a new inode and socket object. The two are bound together |
559 | * and initialised. The socket is then returned. If we are out of inodes | |
560 | * NULL is returned. | |
561 | */ | |
562 | ||
f4a00aac | 563 | struct socket *sock_alloc(void) |
1da177e4 | 564 | { |
89bddce5 SH |
565 | struct inode *inode; |
566 | struct socket *sock; | |
1da177e4 | 567 | |
a209dfc7 | 568 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
569 | if (!inode) |
570 | return NULL; | |
571 | ||
572 | sock = SOCKET_I(inode); | |
573 | ||
85fe4025 | 574 | inode->i_ino = get_next_ino(); |
89bddce5 | 575 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
576 | inode->i_uid = current_fsuid(); |
577 | inode->i_gid = current_fsgid(); | |
600e1779 | 578 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 579 | |
1da177e4 LT |
580 | return sock; |
581 | } | |
f4a00aac | 582 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 583 | |
1da177e4 LT |
584 | /** |
585 | * sock_release - close a socket | |
586 | * @sock: socket to close | |
587 | * | |
588 | * The socket is released from the protocol stack if it has a release | |
589 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 590 | * an inode not a file. |
1da177e4 | 591 | */ |
89bddce5 | 592 | |
1da177e4 LT |
593 | void sock_release(struct socket *sock) |
594 | { | |
595 | if (sock->ops) { | |
596 | struct module *owner = sock->ops->owner; | |
597 | ||
598 | sock->ops->release(sock); | |
599 | sock->ops = NULL; | |
600 | module_put(owner); | |
601 | } | |
602 | ||
eaefd110 | 603 | if (rcu_dereference_protected(sock->wq, 1)->fasync_list) |
3410f22e | 604 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 605 | |
1da177e4 LT |
606 | if (!sock->file) { |
607 | iput(SOCK_INODE(sock)); | |
608 | return; | |
609 | } | |
89bddce5 | 610 | sock->file = NULL; |
1da177e4 | 611 | } |
c6d409cf | 612 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 613 | |
c14ac945 | 614 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 615 | { |
140c55d4 ED |
616 | u8 flags = *tx_flags; |
617 | ||
c14ac945 | 618 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
619 | flags |= SKBTX_HW_TSTAMP; |
620 | ||
c14ac945 | 621 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
622 | flags |= SKBTX_SW_TSTAMP; |
623 | ||
c14ac945 | 624 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
625 | flags |= SKBTX_SCHED_TSTAMP; |
626 | ||
140c55d4 | 627 | *tx_flags = flags; |
20d49473 | 628 | } |
67cc0d40 | 629 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 630 | |
d8725c86 | 631 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 632 | { |
01e97e65 | 633 | int ret = sock->ops->sendmsg(sock, msg, msg_data_left(msg)); |
d8725c86 AV |
634 | BUG_ON(ret == -EIOCBQUEUED); |
635 | return ret; | |
1da177e4 LT |
636 | } |
637 | ||
d8725c86 | 638 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 639 | { |
d8725c86 | 640 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 641 | msg_data_left(msg)); |
228e548e | 642 | |
d8725c86 | 643 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 644 | } |
c6d409cf | 645 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 LT |
646 | |
647 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, | |
648 | struct kvec *vec, size_t num, size_t size) | |
649 | { | |
6aa24814 | 650 | iov_iter_kvec(&msg->msg_iter, WRITE | ITER_KVEC, vec, num, size); |
d8725c86 | 651 | return sock_sendmsg(sock, msg); |
1da177e4 | 652 | } |
c6d409cf | 653 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 654 | |
306b13eb TH |
655 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
656 | struct kvec *vec, size_t num, size_t size) | |
657 | { | |
658 | struct socket *sock = sk->sk_socket; | |
659 | ||
660 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 661 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb TH |
662 | |
663 | iov_iter_kvec(&msg->msg_iter, WRITE | ITER_KVEC, vec, num, size); | |
664 | ||
665 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
666 | } | |
667 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
668 | ||
8605330a SHY |
669 | static bool skb_is_err_queue(const struct sk_buff *skb) |
670 | { | |
671 | /* pkt_type of skbs enqueued on the error queue are set to | |
672 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
673 | * in recvmsg, since skbs received on a local socket will never | |
674 | * have a pkt_type of PACKET_OUTGOING. | |
675 | */ | |
676 | return skb->pkt_type == PACKET_OUTGOING; | |
677 | } | |
678 | ||
b50a5c70 ML |
679 | /* On transmit, software and hardware timestamps are returned independently. |
680 | * As the two skb clones share the hardware timestamp, which may be updated | |
681 | * before the software timestamp is received, a hardware TX timestamp may be | |
682 | * returned only if there is no software TX timestamp. Ignore false software | |
683 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
684 | * option SO_TIMESTAMP(NS) is enabled on the socket, even when the skb has a | |
685 | * hardware timestamp. | |
686 | */ | |
687 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
688 | { | |
689 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
690 | } | |
691 | ||
aad9c8c4 ML |
692 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb) |
693 | { | |
694 | struct scm_ts_pktinfo ts_pktinfo; | |
695 | struct net_device *orig_dev; | |
696 | ||
697 | if (!skb_mac_header_was_set(skb)) | |
698 | return; | |
699 | ||
700 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
701 | ||
702 | rcu_read_lock(); | |
703 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
704 | if (orig_dev) | |
705 | ts_pktinfo.if_index = orig_dev->ifindex; | |
706 | rcu_read_unlock(); | |
707 | ||
708 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
709 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
710 | sizeof(ts_pktinfo), &ts_pktinfo); | |
711 | } | |
712 | ||
92f37fd2 ED |
713 | /* |
714 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
715 | */ | |
716 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
717 | struct sk_buff *skb) | |
718 | { | |
20d49473 | 719 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
f24b9be5 | 720 | struct scm_timestamping tss; |
b50a5c70 | 721 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
722 | struct skb_shared_hwtstamps *shhwtstamps = |
723 | skb_hwtstamps(skb); | |
724 | ||
725 | /* Race occurred between timestamp enabling and packet | |
726 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 727 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 728 | __net_timestamp(skb); |
b50a5c70 ML |
729 | false_tstamp = 1; |
730 | } | |
20d49473 PO |
731 | |
732 | if (need_software_tstamp) { | |
733 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
734 | struct timeval tv; | |
735 | skb_get_timestamp(skb, &tv); | |
736 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMP, | |
737 | sizeof(tv), &tv); | |
738 | } else { | |
f24b9be5 WB |
739 | struct timespec ts; |
740 | skb_get_timestampns(skb, &ts); | |
20d49473 | 741 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPNS, |
f24b9be5 | 742 | sizeof(ts), &ts); |
20d49473 PO |
743 | } |
744 | } | |
745 | ||
f24b9be5 | 746 | memset(&tss, 0, sizeof(tss)); |
c199105d | 747 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
f24b9be5 | 748 | ktime_to_timespec_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 749 | empty = 0; |
4d276eb6 | 750 | if (shhwtstamps && |
b9f40e21 | 751 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
b50a5c70 | 752 | !skb_is_swtx_tstamp(skb, false_tstamp) && |
aad9c8c4 | 753 | ktime_to_timespec_cond(shhwtstamps->hwtstamp, tss.ts + 2)) { |
4d276eb6 | 754 | empty = 0; |
aad9c8c4 ML |
755 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && |
756 | !skb_is_err_queue(skb)) | |
757 | put_ts_pktinfo(msg, skb); | |
758 | } | |
1c885808 | 759 | if (!empty) { |
20d49473 | 760 | put_cmsg(msg, SOL_SOCKET, |
f24b9be5 | 761 | SCM_TIMESTAMPING, sizeof(tss), &tss); |
1c885808 | 762 | |
8605330a | 763 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 764 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
765 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
766 | skb->len, skb->data); | |
767 | } | |
92f37fd2 | 768 | } |
7c81fd8b ACM |
769 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
770 | ||
6e3e939f JB |
771 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
772 | struct sk_buff *skb) | |
773 | { | |
774 | int ack; | |
775 | ||
776 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
777 | return; | |
778 | if (!skb->wifi_acked_valid) | |
779 | return; | |
780 | ||
781 | ack = skb->wifi_acked; | |
782 | ||
783 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
784 | } | |
785 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
786 | ||
11165f14 | 787 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
788 | struct sk_buff *skb) | |
3b885787 | 789 | { |
744d5a3e | 790 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 791 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 792 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
793 | } |
794 | ||
767dd033 | 795 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
796 | struct sk_buff *skb) |
797 | { | |
798 | sock_recv_timestamp(msg, sk, skb); | |
799 | sock_recv_drops(msg, sk, skb); | |
800 | } | |
767dd033 | 801 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 802 | |
1b784140 | 803 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 804 | int flags) |
1da177e4 | 805 | { |
2da62906 | 806 | return sock->ops->recvmsg(sock, msg, msg_data_left(msg), flags); |
1da177e4 LT |
807 | } |
808 | ||
2da62906 | 809 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 810 | { |
2da62906 | 811 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 812 | |
2da62906 | 813 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 814 | } |
c6d409cf | 815 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 816 | |
c1249c0a ML |
817 | /** |
818 | * kernel_recvmsg - Receive a message from a socket (kernel space) | |
819 | * @sock: The socket to receive the message from | |
820 | * @msg: Received message | |
821 | * @vec: Input s/g array for message data | |
822 | * @num: Size of input s/g array | |
823 | * @size: Number of bytes to read | |
824 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
825 | * | |
826 | * On return the msg structure contains the scatter/gather array passed in the | |
827 | * vec argument. The array is modified so that it consists of the unfilled | |
828 | * portion of the original array. | |
829 | * | |
830 | * The returned value is the total number of bytes received, or an error. | |
831 | */ | |
89bddce5 SH |
832 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
833 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
834 | { |
835 | mm_segment_t oldfs = get_fs(); | |
836 | int result; | |
837 | ||
6aa24814 | 838 | iov_iter_kvec(&msg->msg_iter, READ | ITER_KVEC, vec, num, size); |
1da177e4 | 839 | set_fs(KERNEL_DS); |
2da62906 | 840 | result = sock_recvmsg(sock, msg, flags); |
1da177e4 LT |
841 | set_fs(oldfs); |
842 | return result; | |
843 | } | |
c6d409cf | 844 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 845 | |
ce1d4d3e CH |
846 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
847 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 848 | { |
1da177e4 LT |
849 | struct socket *sock; |
850 | int flags; | |
851 | ||
ce1d4d3e CH |
852 | sock = file->private_data; |
853 | ||
35f9c09f ED |
854 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
855 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
856 | flags |= more; | |
ce1d4d3e | 857 | |
e6949583 | 858 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 859 | } |
1da177e4 | 860 | |
9c55e01c | 861 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 862 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
863 | unsigned int flags) |
864 | { | |
865 | struct socket *sock = file->private_data; | |
866 | ||
997b37da RDC |
867 | if (unlikely(!sock->ops->splice_read)) |
868 | return -EINVAL; | |
869 | ||
9c55e01c JA |
870 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
871 | } | |
872 | ||
8ae5e030 | 873 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 874 | { |
6d652330 AV |
875 | struct file *file = iocb->ki_filp; |
876 | struct socket *sock = file->private_data; | |
0345f931 | 877 | struct msghdr msg = {.msg_iter = *to, |
878 | .msg_iocb = iocb}; | |
8ae5e030 | 879 | ssize_t res; |
ce1d4d3e | 880 | |
8ae5e030 AV |
881 | if (file->f_flags & O_NONBLOCK) |
882 | msg.msg_flags = MSG_DONTWAIT; | |
883 | ||
884 | if (iocb->ki_pos != 0) | |
1da177e4 | 885 | return -ESPIPE; |
027445c3 | 886 | |
66ee59af | 887 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
888 | return 0; |
889 | ||
2da62906 | 890 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
891 | *to = msg.msg_iter; |
892 | return res; | |
1da177e4 LT |
893 | } |
894 | ||
8ae5e030 | 895 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 896 | { |
6d652330 AV |
897 | struct file *file = iocb->ki_filp; |
898 | struct socket *sock = file->private_data; | |
0345f931 | 899 | struct msghdr msg = {.msg_iter = *from, |
900 | .msg_iocb = iocb}; | |
8ae5e030 | 901 | ssize_t res; |
1da177e4 | 902 | |
8ae5e030 | 903 | if (iocb->ki_pos != 0) |
ce1d4d3e | 904 | return -ESPIPE; |
027445c3 | 905 | |
8ae5e030 AV |
906 | if (file->f_flags & O_NONBLOCK) |
907 | msg.msg_flags = MSG_DONTWAIT; | |
908 | ||
6d652330 AV |
909 | if (sock->type == SOCK_SEQPACKET) |
910 | msg.msg_flags |= MSG_EOR; | |
911 | ||
d8725c86 | 912 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
913 | *from = msg.msg_iter; |
914 | return res; | |
1da177e4 LT |
915 | } |
916 | ||
1da177e4 LT |
917 | /* |
918 | * Atomic setting of ioctl hooks to avoid race | |
919 | * with module unload. | |
920 | */ | |
921 | ||
4a3e2f71 | 922 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 923 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 924 | |
881d966b | 925 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 926 | { |
4a3e2f71 | 927 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 928 | br_ioctl_hook = hook; |
4a3e2f71 | 929 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
930 | } |
931 | EXPORT_SYMBOL(brioctl_set); | |
932 | ||
4a3e2f71 | 933 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 934 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 935 | |
881d966b | 936 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 937 | { |
4a3e2f71 | 938 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 939 | vlan_ioctl_hook = hook; |
4a3e2f71 | 940 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
941 | } |
942 | EXPORT_SYMBOL(vlan_ioctl_set); | |
943 | ||
4a3e2f71 | 944 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 945 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 946 | |
89bddce5 | 947 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 948 | { |
4a3e2f71 | 949 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 950 | dlci_ioctl_hook = hook; |
4a3e2f71 | 951 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
952 | } |
953 | EXPORT_SYMBOL(dlci_ioctl_set); | |
954 | ||
6b96018b AB |
955 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
956 | unsigned int cmd, unsigned long arg) | |
957 | { | |
958 | int err; | |
959 | void __user *argp = (void __user *)arg; | |
960 | ||
961 | err = sock->ops->ioctl(sock, cmd, arg); | |
962 | ||
963 | /* | |
964 | * If this ioctl is unknown try to hand it down | |
965 | * to the NIC driver. | |
966 | */ | |
36fd633e AV |
967 | if (err != -ENOIOCTLCMD) |
968 | return err; | |
6b96018b | 969 | |
36fd633e AV |
970 | if (cmd == SIOCGIFCONF) { |
971 | struct ifconf ifc; | |
972 | if (copy_from_user(&ifc, argp, sizeof(struct ifconf))) | |
973 | return -EFAULT; | |
974 | rtnl_lock(); | |
975 | err = dev_ifconf(net, &ifc, sizeof(struct ifreq)); | |
976 | rtnl_unlock(); | |
977 | if (!err && copy_to_user(argp, &ifc, sizeof(struct ifconf))) | |
978 | err = -EFAULT; | |
44c02a2c AV |
979 | } else { |
980 | struct ifreq ifr; | |
981 | bool need_copyout; | |
982 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
983 | return -EFAULT; | |
984 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
985 | if (!err && need_copyout) | |
986 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
987 | return -EFAULT; | |
36fd633e | 988 | } |
6b96018b AB |
989 | return err; |
990 | } | |
991 | ||
1da177e4 LT |
992 | /* |
993 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
994 | * what to do with it - that's up to the protocol still. | |
995 | */ | |
996 | ||
d8d211a2 | 997 | struct ns_common *get_net_ns(struct ns_common *ns) |
c62cce2c AV |
998 | { |
999 | return &get_net(container_of(ns, struct net, ns))->ns; | |
1000 | } | |
d8d211a2 | 1001 | EXPORT_SYMBOL_GPL(get_net_ns); |
c62cce2c | 1002 | |
1da177e4 LT |
1003 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
1004 | { | |
1005 | struct socket *sock; | |
881d966b | 1006 | struct sock *sk; |
1da177e4 LT |
1007 | void __user *argp = (void __user *)arg; |
1008 | int pid, err; | |
881d966b | 1009 | struct net *net; |
1da177e4 | 1010 | |
b69aee04 | 1011 | sock = file->private_data; |
881d966b | 1012 | sk = sock->sk; |
3b1e0a65 | 1013 | net = sock_net(sk); |
44c02a2c AV |
1014 | if (unlikely(cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))) { |
1015 | struct ifreq ifr; | |
1016 | bool need_copyout; | |
1017 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
1018 | return -EFAULT; | |
1019 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1020 | if (!err && need_copyout) | |
1021 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
1022 | return -EFAULT; | |
1da177e4 | 1023 | } else |
3d23e349 | 1024 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1025 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
b1b0c245 | 1026 | err = wext_handle_ioctl(net, cmd, argp); |
1da177e4 | 1027 | } else |
3d23e349 | 1028 | #endif |
89bddce5 | 1029 | switch (cmd) { |
1da177e4 LT |
1030 | case FIOSETOWN: |
1031 | case SIOCSPGRP: | |
1032 | err = -EFAULT; | |
1033 | if (get_user(pid, (int __user *)argp)) | |
1034 | break; | |
393cc3f5 | 1035 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1036 | break; |
1037 | case FIOGETOWN: | |
1038 | case SIOCGPGRP: | |
609d7fa9 | 1039 | err = put_user(f_getown(sock->file), |
89bddce5 | 1040 | (int __user *)argp); |
1da177e4 LT |
1041 | break; |
1042 | case SIOCGIFBR: | |
1043 | case SIOCSIFBR: | |
1044 | case SIOCBRADDBR: | |
1045 | case SIOCBRDELBR: | |
1046 | err = -ENOPKG; | |
1047 | if (!br_ioctl_hook) | |
1048 | request_module("bridge"); | |
1049 | ||
4a3e2f71 | 1050 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1051 | if (br_ioctl_hook) |
881d966b | 1052 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1053 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1054 | break; |
1055 | case SIOCGIFVLAN: | |
1056 | case SIOCSIFVLAN: | |
1057 | err = -ENOPKG; | |
1058 | if (!vlan_ioctl_hook) | |
1059 | request_module("8021q"); | |
1060 | ||
4a3e2f71 | 1061 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1062 | if (vlan_ioctl_hook) |
881d966b | 1063 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1064 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1065 | break; |
1da177e4 LT |
1066 | case SIOCADDDLCI: |
1067 | case SIOCDELDLCI: | |
1068 | err = -ENOPKG; | |
1069 | if (!dlci_ioctl_hook) | |
1070 | request_module("dlci"); | |
1071 | ||
7512cbf6 PE |
1072 | mutex_lock(&dlci_ioctl_mutex); |
1073 | if (dlci_ioctl_hook) | |
1da177e4 | 1074 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1075 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 | 1076 | break; |
c62cce2c AV |
1077 | case SIOCGSKNS: |
1078 | err = -EPERM; | |
1079 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1080 | break; | |
1081 | ||
1082 | err = open_related_ns(&net->ns, get_net_ns); | |
1083 | break; | |
1da177e4 | 1084 | default: |
6b96018b | 1085 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1086 | break; |
89bddce5 | 1087 | } |
1da177e4 LT |
1088 | return err; |
1089 | } | |
1090 | ||
1091 | int sock_create_lite(int family, int type, int protocol, struct socket **res) | |
1092 | { | |
1093 | int err; | |
1094 | struct socket *sock = NULL; | |
89bddce5 | 1095 | |
1da177e4 LT |
1096 | err = security_socket_create(family, type, protocol, 1); |
1097 | if (err) | |
1098 | goto out; | |
1099 | ||
1100 | sock = sock_alloc(); | |
1101 | if (!sock) { | |
1102 | err = -ENOMEM; | |
1103 | goto out; | |
1104 | } | |
1105 | ||
1da177e4 | 1106 | sock->type = type; |
7420ed23 VY |
1107 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1108 | if (err) | |
1109 | goto out_release; | |
1110 | ||
1da177e4 LT |
1111 | out: |
1112 | *res = sock; | |
1113 | return err; | |
7420ed23 VY |
1114 | out_release: |
1115 | sock_release(sock); | |
1116 | sock = NULL; | |
1117 | goto out; | |
1da177e4 | 1118 | } |
c6d409cf | 1119 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 | 1120 | |
15252423 CH |
1121 | static struct wait_queue_head *sock_get_poll_head(struct file *file, |
1122 | __poll_t events) | |
1123 | { | |
1124 | struct socket *sock = file->private_data; | |
1125 | ||
1126 | if (!sock->ops->poll_mask) | |
1127 | return NULL; | |
1128 | sock_poll_busy_loop(sock, events); | |
1129 | return sk_sleep(sock->sk); | |
1130 | } | |
1131 | ||
1132 | static __poll_t sock_poll_mask(struct file *file, __poll_t events) | |
1133 | { | |
1134 | struct socket *sock = file->private_data; | |
1135 | ||
1136 | /* | |
1137 | * We need to be sure we are in sync with the socket flags modification. | |
1138 | * | |
1139 | * This memory barrier is paired in the wq_has_sleeper. | |
1140 | */ | |
1141 | smp_mb(); | |
1142 | ||
1143 | /* this socket can poll_ll so tell the system call */ | |
1144 | return sock->ops->poll_mask(sock, events) | | |
1145 | (sk_can_busy_loop(sock->sk) ? POLL_BUSY_LOOP : 0); | |
1146 | } | |
1147 | ||
1da177e4 | 1148 | /* No kernel lock held - perfect */ |
ade994f4 | 1149 | static __poll_t sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1150 | { |
3cafb376 | 1151 | struct socket *sock = file->private_data; |
15252423 CH |
1152 | __poll_t events = poll_requested_events(wait), mask = 0; |
1153 | ||
1154 | if (sock->ops->poll) { | |
1155 | sock_poll_busy_loop(sock, events); | |
1156 | mask = sock->ops->poll(file, sock, wait); | |
1157 | } else if (sock->ops->poll_mask) { | |
1158 | sock_poll_wait(file, sock_get_poll_head(file, events), wait); | |
1159 | mask = sock->ops->poll_mask(sock, events); | |
1160 | } | |
2d48d67f | 1161 | |
15252423 | 1162 | return mask | sock_poll_busy_flag(sock); |
1da177e4 LT |
1163 | } |
1164 | ||
89bddce5 | 1165 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1166 | { |
b69aee04 | 1167 | struct socket *sock = file->private_data; |
1da177e4 LT |
1168 | |
1169 | return sock->ops->mmap(file, sock, vma); | |
1170 | } | |
1171 | ||
20380731 | 1172 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1173 | { |
1da177e4 LT |
1174 | sock_release(SOCKET_I(inode)); |
1175 | return 0; | |
1176 | } | |
1177 | ||
1178 | /* | |
1179 | * Update the socket async list | |
1180 | * | |
1181 | * Fasync_list locking strategy. | |
1182 | * | |
1183 | * 1. fasync_list is modified only under process context socket lock | |
1184 | * i.e. under semaphore. | |
1185 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1186 | * or under socket lock |
1da177e4 LT |
1187 | */ |
1188 | ||
1189 | static int sock_fasync(int fd, struct file *filp, int on) | |
1190 | { | |
989a2979 ED |
1191 | struct socket *sock = filp->private_data; |
1192 | struct sock *sk = sock->sk; | |
eaefd110 | 1193 | struct socket_wq *wq; |
1da177e4 | 1194 | |
989a2979 | 1195 | if (sk == NULL) |
1da177e4 | 1196 | return -EINVAL; |
1da177e4 LT |
1197 | |
1198 | lock_sock(sk); | |
1e1d04e6 | 1199 | wq = rcu_dereference_protected(sock->wq, lockdep_sock_is_held(sk)); |
eaefd110 | 1200 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1201 | |
eaefd110 | 1202 | if (!wq->fasync_list) |
989a2979 ED |
1203 | sock_reset_flag(sk, SOCK_FASYNC); |
1204 | else | |
bcdce719 | 1205 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1206 | |
989a2979 | 1207 | release_sock(sk); |
1da177e4 LT |
1208 | return 0; |
1209 | } | |
1210 | ||
ceb5d58b | 1211 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1212 | |
ceb5d58b | 1213 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1214 | { |
ceb5d58b | 1215 | if (!wq || !wq->fasync_list) |
1da177e4 | 1216 | return -1; |
ceb5d58b | 1217 | |
89bddce5 | 1218 | switch (how) { |
8d8ad9d7 | 1219 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1220 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1221 | break; |
1222 | goto call_kill; | |
8d8ad9d7 | 1223 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1224 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 LT |
1225 | break; |
1226 | /* fall through */ | |
8d8ad9d7 | 1227 | case SOCK_WAKE_IO: |
89bddce5 | 1228 | call_kill: |
43815482 | 1229 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1230 | break; |
8d8ad9d7 | 1231 | case SOCK_WAKE_URG: |
43815482 | 1232 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1233 | } |
ceb5d58b | 1234 | |
1da177e4 LT |
1235 | return 0; |
1236 | } | |
c6d409cf | 1237 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1238 | |
721db93a | 1239 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1240 | struct socket **res, int kern) |
1da177e4 LT |
1241 | { |
1242 | int err; | |
1243 | struct socket *sock; | |
55737fda | 1244 | const struct net_proto_family *pf; |
1da177e4 LT |
1245 | |
1246 | /* | |
89bddce5 | 1247 | * Check protocol is in range |
1da177e4 LT |
1248 | */ |
1249 | if (family < 0 || family >= NPROTO) | |
1250 | return -EAFNOSUPPORT; | |
1251 | if (type < 0 || type >= SOCK_MAX) | |
1252 | return -EINVAL; | |
1253 | ||
1254 | /* Compatibility. | |
1255 | ||
1256 | This uglymoron is moved from INET layer to here to avoid | |
1257 | deadlock in module load. | |
1258 | */ | |
1259 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1260 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1261 | current->comm); | |
1da177e4 LT |
1262 | family = PF_PACKET; |
1263 | } | |
1264 | ||
1265 | err = security_socket_create(family, type, protocol, kern); | |
1266 | if (err) | |
1267 | return err; | |
89bddce5 | 1268 | |
55737fda SH |
1269 | /* |
1270 | * Allocate the socket and allow the family to set things up. if | |
1271 | * the protocol is 0, the family is instructed to select an appropriate | |
1272 | * default. | |
1273 | */ | |
1274 | sock = sock_alloc(); | |
1275 | if (!sock) { | |
e87cc472 | 1276 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1277 | return -ENFILE; /* Not exactly a match, but its the |
1278 | closest posix thing */ | |
1279 | } | |
1280 | ||
1281 | sock->type = type; | |
1282 | ||
95a5afca | 1283 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1284 | /* Attempt to load a protocol module if the find failed. |
1285 | * | |
1286 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1287 | * requested real, full-featured networking support upon configuration. |
1288 | * Otherwise module support will break! | |
1289 | */ | |
190683a9 | 1290 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1291 | request_module("net-pf-%d", family); |
1da177e4 LT |
1292 | #endif |
1293 | ||
55737fda SH |
1294 | rcu_read_lock(); |
1295 | pf = rcu_dereference(net_families[family]); | |
1296 | err = -EAFNOSUPPORT; | |
1297 | if (!pf) | |
1298 | goto out_release; | |
1da177e4 LT |
1299 | |
1300 | /* | |
1301 | * We will call the ->create function, that possibly is in a loadable | |
1302 | * module, so we have to bump that loadable module refcnt first. | |
1303 | */ | |
55737fda | 1304 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1305 | goto out_release; |
1306 | ||
55737fda SH |
1307 | /* Now protected by module ref count */ |
1308 | rcu_read_unlock(); | |
1309 | ||
3f378b68 | 1310 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1311 | if (err < 0) |
1da177e4 | 1312 | goto out_module_put; |
a79af59e | 1313 | |
1da177e4 LT |
1314 | /* |
1315 | * Now to bump the refcnt of the [loadable] module that owns this | |
1316 | * socket at sock_release time we decrement its refcnt. | |
1317 | */ | |
55737fda SH |
1318 | if (!try_module_get(sock->ops->owner)) |
1319 | goto out_module_busy; | |
1320 | ||
1da177e4 LT |
1321 | /* |
1322 | * Now that we're done with the ->create function, the [loadable] | |
1323 | * module can have its refcnt decremented | |
1324 | */ | |
55737fda | 1325 | module_put(pf->owner); |
7420ed23 VY |
1326 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1327 | if (err) | |
3b185525 | 1328 | goto out_sock_release; |
55737fda | 1329 | *res = sock; |
1da177e4 | 1330 | |
55737fda SH |
1331 | return 0; |
1332 | ||
1333 | out_module_busy: | |
1334 | err = -EAFNOSUPPORT; | |
1da177e4 | 1335 | out_module_put: |
55737fda SH |
1336 | sock->ops = NULL; |
1337 | module_put(pf->owner); | |
1338 | out_sock_release: | |
1da177e4 | 1339 | sock_release(sock); |
55737fda SH |
1340 | return err; |
1341 | ||
1342 | out_release: | |
1343 | rcu_read_unlock(); | |
1344 | goto out_sock_release; | |
1da177e4 | 1345 | } |
721db93a | 1346 | EXPORT_SYMBOL(__sock_create); |
1da177e4 LT |
1347 | |
1348 | int sock_create(int family, int type, int protocol, struct socket **res) | |
1349 | { | |
1b8d7ae4 | 1350 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1351 | } |
c6d409cf | 1352 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1353 | |
eeb1bd5c | 1354 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1355 | { |
eeb1bd5c | 1356 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1357 | } |
c6d409cf | 1358 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1359 | |
9d6a15c3 | 1360 | int __sys_socket(int family, int type, int protocol) |
1da177e4 LT |
1361 | { |
1362 | int retval; | |
1363 | struct socket *sock; | |
a677a039 UD |
1364 | int flags; |
1365 | ||
e38b36f3 UD |
1366 | /* Check the SOCK_* constants for consistency. */ |
1367 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1368 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1369 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1370 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1371 | ||
a677a039 | 1372 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1373 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1374 | return -EINVAL; |
1375 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1376 | |
aaca0bdc UD |
1377 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1378 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1379 | ||
1da177e4 LT |
1380 | retval = sock_create(family, type, protocol, &sock); |
1381 | if (retval < 0) | |
8e1611e2 | 1382 | return retval; |
1da177e4 | 1383 | |
8e1611e2 | 1384 | return sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1385 | } |
1386 | ||
9d6a15c3 DB |
1387 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1388 | { | |
1389 | return __sys_socket(family, type, protocol); | |
1390 | } | |
1391 | ||
1da177e4 LT |
1392 | /* |
1393 | * Create a pair of connected sockets. | |
1394 | */ | |
1395 | ||
6debc8d8 | 1396 | int __sys_socketpair(int family, int type, int protocol, int __user *usockvec) |
1da177e4 LT |
1397 | { |
1398 | struct socket *sock1, *sock2; | |
1399 | int fd1, fd2, err; | |
db349509 | 1400 | struct file *newfile1, *newfile2; |
a677a039 UD |
1401 | int flags; |
1402 | ||
1403 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1404 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1405 | return -EINVAL; |
1406 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1407 | |
aaca0bdc UD |
1408 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1409 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1410 | ||
016a266b AV |
1411 | /* |
1412 | * reserve descriptors and make sure we won't fail | |
1413 | * to return them to userland. | |
1414 | */ | |
1415 | fd1 = get_unused_fd_flags(flags); | |
1416 | if (unlikely(fd1 < 0)) | |
1417 | return fd1; | |
1418 | ||
1419 | fd2 = get_unused_fd_flags(flags); | |
1420 | if (unlikely(fd2 < 0)) { | |
1421 | put_unused_fd(fd1); | |
1422 | return fd2; | |
1423 | } | |
1424 | ||
1425 | err = put_user(fd1, &usockvec[0]); | |
1426 | if (err) | |
1427 | goto out; | |
1428 | ||
1429 | err = put_user(fd2, &usockvec[1]); | |
1430 | if (err) | |
1431 | goto out; | |
1432 | ||
1da177e4 LT |
1433 | /* |
1434 | * Obtain the first socket and check if the underlying protocol | |
1435 | * supports the socketpair call. | |
1436 | */ | |
1437 | ||
1438 | err = sock_create(family, type, protocol, &sock1); | |
016a266b | 1439 | if (unlikely(err < 0)) |
1da177e4 LT |
1440 | goto out; |
1441 | ||
1442 | err = sock_create(family, type, protocol, &sock2); | |
016a266b AV |
1443 | if (unlikely(err < 0)) { |
1444 | sock_release(sock1); | |
1445 | goto out; | |
bf3c23d1 | 1446 | } |
d73aa286 | 1447 | |
d47cd945 DH |
1448 | err = security_socket_socketpair(sock1, sock2); |
1449 | if (unlikely(err)) { | |
1450 | sock_release(sock2); | |
1451 | sock_release(sock1); | |
1452 | goto out; | |
1453 | } | |
1454 | ||
016a266b AV |
1455 | err = sock1->ops->socketpair(sock1, sock2); |
1456 | if (unlikely(err < 0)) { | |
1457 | sock_release(sock2); | |
1458 | sock_release(sock1); | |
1459 | goto out; | |
28407630 AV |
1460 | } |
1461 | ||
aab174f0 | 1462 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1463 | if (IS_ERR(newfile1)) { |
28407630 | 1464 | err = PTR_ERR(newfile1); |
016a266b AV |
1465 | sock_release(sock2); |
1466 | goto out; | |
28407630 AV |
1467 | } |
1468 | ||
aab174f0 | 1469 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1470 | if (IS_ERR(newfile2)) { |
1471 | err = PTR_ERR(newfile2); | |
016a266b AV |
1472 | fput(newfile1); |
1473 | goto out; | |
db349509 AV |
1474 | } |
1475 | ||
157cf649 | 1476 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1477 | |
db349509 AV |
1478 | fd_install(fd1, newfile1); |
1479 | fd_install(fd2, newfile2); | |
d73aa286 | 1480 | return 0; |
1da177e4 | 1481 | |
016a266b | 1482 | out: |
d73aa286 | 1483 | put_unused_fd(fd2); |
d73aa286 | 1484 | put_unused_fd(fd1); |
1da177e4 LT |
1485 | return err; |
1486 | } | |
1487 | ||
6debc8d8 DB |
1488 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1489 | int __user *, usockvec) | |
1490 | { | |
1491 | return __sys_socketpair(family, type, protocol, usockvec); | |
1492 | } | |
1493 | ||
1da177e4 LT |
1494 | /* |
1495 | * Bind a name to a socket. Nothing much to do here since it's | |
1496 | * the protocol's responsibility to handle the local address. | |
1497 | * | |
1498 | * We move the socket address to kernel space before we call | |
1499 | * the protocol layer (having also checked the address is ok). | |
1500 | */ | |
1501 | ||
a87d35d8 | 1502 | int __sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
1da177e4 LT |
1503 | { |
1504 | struct socket *sock; | |
230b1839 | 1505 | struct sockaddr_storage address; |
6cb153ca | 1506 | int err, fput_needed; |
1da177e4 | 1507 | |
89bddce5 | 1508 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1509 | if (sock) { |
43db362d | 1510 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
89bddce5 SH |
1511 | if (err >= 0) { |
1512 | err = security_socket_bind(sock, | |
230b1839 | 1513 | (struct sockaddr *)&address, |
89bddce5 | 1514 | addrlen); |
6cb153ca BL |
1515 | if (!err) |
1516 | err = sock->ops->bind(sock, | |
89bddce5 | 1517 | (struct sockaddr *) |
230b1839 | 1518 | &address, addrlen); |
1da177e4 | 1519 | } |
6cb153ca | 1520 | fput_light(sock->file, fput_needed); |
89bddce5 | 1521 | } |
1da177e4 LT |
1522 | return err; |
1523 | } | |
1524 | ||
a87d35d8 DB |
1525 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1526 | { | |
1527 | return __sys_bind(fd, umyaddr, addrlen); | |
1528 | } | |
1529 | ||
1da177e4 LT |
1530 | /* |
1531 | * Perform a listen. Basically, we allow the protocol to do anything | |
1532 | * necessary for a listen, and if that works, we mark the socket as | |
1533 | * ready for listening. | |
1534 | */ | |
1535 | ||
25e290ee | 1536 | int __sys_listen(int fd, int backlog) |
1da177e4 LT |
1537 | { |
1538 | struct socket *sock; | |
6cb153ca | 1539 | int err, fput_needed; |
b8e1f9b5 | 1540 | int somaxconn; |
89bddce5 SH |
1541 | |
1542 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1543 | if (sock) { | |
8efa6e93 | 1544 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1545 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1546 | backlog = somaxconn; |
1da177e4 LT |
1547 | |
1548 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1549 | if (!err) |
1550 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1551 | |
6cb153ca | 1552 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1553 | } |
1554 | return err; | |
1555 | } | |
1556 | ||
25e290ee DB |
1557 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1558 | { | |
1559 | return __sys_listen(fd, backlog); | |
1560 | } | |
1561 | ||
1da177e4 LT |
1562 | /* |
1563 | * For accept, we attempt to create a new socket, set up the link | |
1564 | * with the client, wake up the client, then return the new | |
1565 | * connected fd. We collect the address of the connector in kernel | |
1566 | * space and move it to user at the very end. This is unclean because | |
1567 | * we open the socket then return an error. | |
1568 | * | |
1569 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1570 | * status to recvmsg. We need to add that support in a way thats | |
b903036a | 1571 | * clean when we restructure accept also. |
1da177e4 LT |
1572 | */ |
1573 | ||
4541e805 DB |
1574 | int __sys_accept4(int fd, struct sockaddr __user *upeer_sockaddr, |
1575 | int __user *upeer_addrlen, int flags) | |
1da177e4 LT |
1576 | { |
1577 | struct socket *sock, *newsock; | |
39d8c1b6 | 1578 | struct file *newfile; |
6cb153ca | 1579 | int err, len, newfd, fput_needed; |
230b1839 | 1580 | struct sockaddr_storage address; |
1da177e4 | 1581 | |
77d27200 | 1582 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1583 | return -EINVAL; |
1584 | ||
1585 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1586 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1587 | ||
6cb153ca | 1588 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1589 | if (!sock) |
1590 | goto out; | |
1591 | ||
1592 | err = -ENFILE; | |
c6d409cf ED |
1593 | newsock = sock_alloc(); |
1594 | if (!newsock) | |
1da177e4 LT |
1595 | goto out_put; |
1596 | ||
1597 | newsock->type = sock->type; | |
1598 | newsock->ops = sock->ops; | |
1599 | ||
1da177e4 LT |
1600 | /* |
1601 | * We don't need try_module_get here, as the listening socket (sock) | |
1602 | * has the protocol module (sock->ops->owner) held. | |
1603 | */ | |
1604 | __module_get(newsock->ops->owner); | |
1605 | ||
28407630 | 1606 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1607 | if (unlikely(newfd < 0)) { |
1608 | err = newfd; | |
9a1875e6 DM |
1609 | sock_release(newsock); |
1610 | goto out_put; | |
39d8c1b6 | 1611 | } |
aab174f0 | 1612 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1613 | if (IS_ERR(newfile)) { |
28407630 AV |
1614 | err = PTR_ERR(newfile); |
1615 | put_unused_fd(newfd); | |
28407630 AV |
1616 | goto out_put; |
1617 | } | |
39d8c1b6 | 1618 | |
a79af59e FF |
1619 | err = security_socket_accept(sock, newsock); |
1620 | if (err) | |
39d8c1b6 | 1621 | goto out_fd; |
a79af59e | 1622 | |
cdfbabfb | 1623 | err = sock->ops->accept(sock, newsock, sock->file->f_flags, false); |
1da177e4 | 1624 | if (err < 0) |
39d8c1b6 | 1625 | goto out_fd; |
1da177e4 LT |
1626 | |
1627 | if (upeer_sockaddr) { | |
9b2c45d4 DV |
1628 | len = newsock->ops->getname(newsock, |
1629 | (struct sockaddr *)&address, 2); | |
1630 | if (len < 0) { | |
1da177e4 | 1631 | err = -ECONNABORTED; |
39d8c1b6 | 1632 | goto out_fd; |
1da177e4 | 1633 | } |
43db362d | 1634 | err = move_addr_to_user(&address, |
230b1839 | 1635 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1636 | if (err < 0) |
39d8c1b6 | 1637 | goto out_fd; |
1da177e4 LT |
1638 | } |
1639 | ||
1640 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1641 | ||
39d8c1b6 DM |
1642 | fd_install(newfd, newfile); |
1643 | err = newfd; | |
1da177e4 | 1644 | |
1da177e4 | 1645 | out_put: |
6cb153ca | 1646 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1647 | out: |
1648 | return err; | |
39d8c1b6 | 1649 | out_fd: |
9606a216 | 1650 | fput(newfile); |
39d8c1b6 | 1651 | put_unused_fd(newfd); |
1da177e4 LT |
1652 | goto out_put; |
1653 | } | |
1654 | ||
4541e805 DB |
1655 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1656 | int __user *, upeer_addrlen, int, flags) | |
1657 | { | |
1658 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, flags); | |
1659 | } | |
1660 | ||
20f37034 HC |
1661 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1662 | int __user *, upeer_addrlen) | |
aaca0bdc | 1663 | { |
4541e805 | 1664 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1665 | } |
1666 | ||
1da177e4 LT |
1667 | /* |
1668 | * Attempt to connect to a socket with the server address. The address | |
1669 | * is in user space so we verify it is OK and move it to kernel space. | |
1670 | * | |
1671 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1672 | * break bindings | |
1673 | * | |
1674 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1675 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1676 | * include the -EINPROGRESS status for such sockets. | |
1677 | */ | |
1678 | ||
1387c2c2 | 1679 | int __sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
1da177e4 LT |
1680 | { |
1681 | struct socket *sock; | |
230b1839 | 1682 | struct sockaddr_storage address; |
6cb153ca | 1683 | int err, fput_needed; |
1da177e4 | 1684 | |
6cb153ca | 1685 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1686 | if (!sock) |
1687 | goto out; | |
43db362d | 1688 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1689 | if (err < 0) |
1690 | goto out_put; | |
1691 | ||
89bddce5 | 1692 | err = |
230b1839 | 1693 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1694 | if (err) |
1695 | goto out_put; | |
1696 | ||
230b1839 | 1697 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1698 | sock->file->f_flags); |
1699 | out_put: | |
6cb153ca | 1700 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1701 | out: |
1702 | return err; | |
1703 | } | |
1704 | ||
1387c2c2 DB |
1705 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1706 | int, addrlen) | |
1707 | { | |
1708 | return __sys_connect(fd, uservaddr, addrlen); | |
1709 | } | |
1710 | ||
1da177e4 LT |
1711 | /* |
1712 | * Get the local address ('name') of a socket object. Move the obtained | |
1713 | * name to user space. | |
1714 | */ | |
1715 | ||
8882a107 DB |
1716 | int __sys_getsockname(int fd, struct sockaddr __user *usockaddr, |
1717 | int __user *usockaddr_len) | |
1da177e4 LT |
1718 | { |
1719 | struct socket *sock; | |
230b1839 | 1720 | struct sockaddr_storage address; |
9b2c45d4 | 1721 | int err, fput_needed; |
89bddce5 | 1722 | |
6cb153ca | 1723 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1724 | if (!sock) |
1725 | goto out; | |
1726 | ||
1727 | err = security_socket_getsockname(sock); | |
1728 | if (err) | |
1729 | goto out_put; | |
1730 | ||
9b2c45d4 DV |
1731 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 0); |
1732 | if (err < 0) | |
1da177e4 | 1733 | goto out_put; |
9b2c45d4 DV |
1734 | /* "err" is actually length in this case */ |
1735 | err = move_addr_to_user(&address, err, usockaddr, usockaddr_len); | |
1da177e4 LT |
1736 | |
1737 | out_put: | |
6cb153ca | 1738 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1739 | out: |
1740 | return err; | |
1741 | } | |
1742 | ||
8882a107 DB |
1743 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1744 | int __user *, usockaddr_len) | |
1745 | { | |
1746 | return __sys_getsockname(fd, usockaddr, usockaddr_len); | |
1747 | } | |
1748 | ||
1da177e4 LT |
1749 | /* |
1750 | * Get the remote address ('name') of a socket object. Move the obtained | |
1751 | * name to user space. | |
1752 | */ | |
1753 | ||
b21c8f83 DB |
1754 | int __sys_getpeername(int fd, struct sockaddr __user *usockaddr, |
1755 | int __user *usockaddr_len) | |
1da177e4 LT |
1756 | { |
1757 | struct socket *sock; | |
230b1839 | 1758 | struct sockaddr_storage address; |
9b2c45d4 | 1759 | int err, fput_needed; |
1da177e4 | 1760 | |
89bddce5 SH |
1761 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1762 | if (sock != NULL) { | |
1da177e4 LT |
1763 | err = security_socket_getpeername(sock); |
1764 | if (err) { | |
6cb153ca | 1765 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1766 | return err; |
1767 | } | |
1768 | ||
9b2c45d4 DV |
1769 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 1); |
1770 | if (err >= 0) | |
1771 | /* "err" is actually length in this case */ | |
1772 | err = move_addr_to_user(&address, err, usockaddr, | |
89bddce5 | 1773 | usockaddr_len); |
6cb153ca | 1774 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1775 | } |
1776 | return err; | |
1777 | } | |
1778 | ||
b21c8f83 DB |
1779 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1780 | int __user *, usockaddr_len) | |
1781 | { | |
1782 | return __sys_getpeername(fd, usockaddr, usockaddr_len); | |
1783 | } | |
1784 | ||
1da177e4 LT |
1785 | /* |
1786 | * Send a datagram to a given address. We move the address into kernel | |
1787 | * space and check the user space data area is readable before invoking | |
1788 | * the protocol. | |
1789 | */ | |
211b634b DB |
1790 | int __sys_sendto(int fd, void __user *buff, size_t len, unsigned int flags, |
1791 | struct sockaddr __user *addr, int addr_len) | |
1da177e4 LT |
1792 | { |
1793 | struct socket *sock; | |
230b1839 | 1794 | struct sockaddr_storage address; |
1da177e4 LT |
1795 | int err; |
1796 | struct msghdr msg; | |
1797 | struct iovec iov; | |
6cb153ca | 1798 | int fput_needed; |
6cb153ca | 1799 | |
602bd0e9 AV |
1800 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1801 | if (unlikely(err)) | |
1802 | return err; | |
de0fa95c PE |
1803 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1804 | if (!sock) | |
4387ff75 | 1805 | goto out; |
6cb153ca | 1806 | |
89bddce5 | 1807 | msg.msg_name = NULL; |
89bddce5 SH |
1808 | msg.msg_control = NULL; |
1809 | msg.msg_controllen = 0; | |
1810 | msg.msg_namelen = 0; | |
6cb153ca | 1811 | if (addr) { |
43db362d | 1812 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1813 | if (err < 0) |
1814 | goto out_put; | |
230b1839 | 1815 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1816 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1817 | } |
1818 | if (sock->file->f_flags & O_NONBLOCK) | |
1819 | flags |= MSG_DONTWAIT; | |
1820 | msg.msg_flags = flags; | |
d8725c86 | 1821 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 1822 | |
89bddce5 | 1823 | out_put: |
de0fa95c | 1824 | fput_light(sock->file, fput_needed); |
4387ff75 | 1825 | out: |
1da177e4 LT |
1826 | return err; |
1827 | } | |
1828 | ||
211b634b DB |
1829 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
1830 | unsigned int, flags, struct sockaddr __user *, addr, | |
1831 | int, addr_len) | |
1832 | { | |
1833 | return __sys_sendto(fd, buff, len, flags, addr, addr_len); | |
1834 | } | |
1835 | ||
1da177e4 | 1836 | /* |
89bddce5 | 1837 | * Send a datagram down a socket. |
1da177e4 LT |
1838 | */ |
1839 | ||
3e0fa65f | 1840 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1841 | unsigned int, flags) |
1da177e4 | 1842 | { |
211b634b | 1843 | return __sys_sendto(fd, buff, len, flags, NULL, 0); |
1da177e4 LT |
1844 | } |
1845 | ||
1846 | /* | |
89bddce5 | 1847 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1848 | * sender. We verify the buffers are writable and if needed move the |
1849 | * sender address from kernel to user space. | |
1850 | */ | |
7a09e1eb DB |
1851 | int __sys_recvfrom(int fd, void __user *ubuf, size_t size, unsigned int flags, |
1852 | struct sockaddr __user *addr, int __user *addr_len) | |
1da177e4 LT |
1853 | { |
1854 | struct socket *sock; | |
1855 | struct iovec iov; | |
1856 | struct msghdr msg; | |
230b1839 | 1857 | struct sockaddr_storage address; |
89bddce5 | 1858 | int err, err2; |
6cb153ca BL |
1859 | int fput_needed; |
1860 | ||
602bd0e9 AV |
1861 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
1862 | if (unlikely(err)) | |
1863 | return err; | |
de0fa95c | 1864 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 1865 | if (!sock) |
de0fa95c | 1866 | goto out; |
1da177e4 | 1867 | |
89bddce5 SH |
1868 | msg.msg_control = NULL; |
1869 | msg.msg_controllen = 0; | |
f3d33426 HFS |
1870 | /* Save some cycles and don't copy the address if not needed */ |
1871 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
1872 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
1873 | msg.msg_namelen = 0; | |
130ed5d1 | 1874 | msg.msg_iocb = NULL; |
9f138fa6 | 1875 | msg.msg_flags = 0; |
1da177e4 LT |
1876 | if (sock->file->f_flags & O_NONBLOCK) |
1877 | flags |= MSG_DONTWAIT; | |
2da62906 | 1878 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 1879 | |
89bddce5 | 1880 | if (err >= 0 && addr != NULL) { |
43db362d | 1881 | err2 = move_addr_to_user(&address, |
230b1839 | 1882 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
1883 | if (err2 < 0) |
1884 | err = err2; | |
1da177e4 | 1885 | } |
de0fa95c PE |
1886 | |
1887 | fput_light(sock->file, fput_needed); | |
4387ff75 | 1888 | out: |
1da177e4 LT |
1889 | return err; |
1890 | } | |
1891 | ||
7a09e1eb DB |
1892 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
1893 | unsigned int, flags, struct sockaddr __user *, addr, | |
1894 | int __user *, addr_len) | |
1895 | { | |
1896 | return __sys_recvfrom(fd, ubuf, size, flags, addr, addr_len); | |
1897 | } | |
1898 | ||
1da177e4 | 1899 | /* |
89bddce5 | 1900 | * Receive a datagram from a socket. |
1da177e4 LT |
1901 | */ |
1902 | ||
b7c0ddf5 JG |
1903 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
1904 | unsigned int, flags) | |
1da177e4 | 1905 | { |
7a09e1eb | 1906 | return __sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
1da177e4 LT |
1907 | } |
1908 | ||
1909 | /* | |
1910 | * Set a socket option. Because we don't know the option lengths we have | |
1911 | * to pass the user mode parameter for the protocols to sort out. | |
1912 | */ | |
1913 | ||
cc36dca0 DB |
1914 | static int __sys_setsockopt(int fd, int level, int optname, |
1915 | char __user *optval, int optlen) | |
1da177e4 | 1916 | { |
6cb153ca | 1917 | int err, fput_needed; |
1da177e4 LT |
1918 | struct socket *sock; |
1919 | ||
1920 | if (optlen < 0) | |
1921 | return -EINVAL; | |
89bddce5 SH |
1922 | |
1923 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1924 | if (sock != NULL) { | |
1925 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
1926 | if (err) |
1927 | goto out_put; | |
1da177e4 LT |
1928 | |
1929 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1930 | err = |
1931 | sock_setsockopt(sock, level, optname, optval, | |
1932 | optlen); | |
1da177e4 | 1933 | else |
89bddce5 SH |
1934 | err = |
1935 | sock->ops->setsockopt(sock, level, optname, optval, | |
1936 | optlen); | |
6cb153ca BL |
1937 | out_put: |
1938 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1939 | } |
1940 | return err; | |
1941 | } | |
1942 | ||
cc36dca0 DB |
1943 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
1944 | char __user *, optval, int, optlen) | |
1945 | { | |
1946 | return __sys_setsockopt(fd, level, optname, optval, optlen); | |
1947 | } | |
1948 | ||
1da177e4 LT |
1949 | /* |
1950 | * Get a socket option. Because we don't know the option lengths we have | |
1951 | * to pass a user mode parameter for the protocols to sort out. | |
1952 | */ | |
1953 | ||
13a2d70e DB |
1954 | static int __sys_getsockopt(int fd, int level, int optname, |
1955 | char __user *optval, int __user *optlen) | |
1da177e4 | 1956 | { |
6cb153ca | 1957 | int err, fput_needed; |
1da177e4 LT |
1958 | struct socket *sock; |
1959 | ||
89bddce5 SH |
1960 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1961 | if (sock != NULL) { | |
6cb153ca BL |
1962 | err = security_socket_getsockopt(sock, level, optname); |
1963 | if (err) | |
1964 | goto out_put; | |
1da177e4 LT |
1965 | |
1966 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1967 | err = |
1968 | sock_getsockopt(sock, level, optname, optval, | |
1969 | optlen); | |
1da177e4 | 1970 | else |
89bddce5 SH |
1971 | err = |
1972 | sock->ops->getsockopt(sock, level, optname, optval, | |
1973 | optlen); | |
6cb153ca BL |
1974 | out_put: |
1975 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1976 | } |
1977 | return err; | |
1978 | } | |
1979 | ||
13a2d70e DB |
1980 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
1981 | char __user *, optval, int __user *, optlen) | |
1982 | { | |
1983 | return __sys_getsockopt(fd, level, optname, optval, optlen); | |
1984 | } | |
1985 | ||
1da177e4 LT |
1986 | /* |
1987 | * Shutdown a socket. | |
1988 | */ | |
1989 | ||
005a1aea | 1990 | int __sys_shutdown(int fd, int how) |
1da177e4 | 1991 | { |
6cb153ca | 1992 | int err, fput_needed; |
1da177e4 LT |
1993 | struct socket *sock; |
1994 | ||
89bddce5 SH |
1995 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1996 | if (sock != NULL) { | |
1da177e4 | 1997 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
1998 | if (!err) |
1999 | err = sock->ops->shutdown(sock, how); | |
2000 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2001 | } |
2002 | return err; | |
2003 | } | |
2004 | ||
005a1aea DB |
2005 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
2006 | { | |
2007 | return __sys_shutdown(fd, how); | |
2008 | } | |
2009 | ||
89bddce5 | 2010 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
2011 | * fields which are the same type (int / unsigned) on our platforms. |
2012 | */ | |
2013 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
2014 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
2015 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
2016 | ||
c71d8ebe TH |
2017 | struct used_address { |
2018 | struct sockaddr_storage name; | |
2019 | unsigned int name_len; | |
2020 | }; | |
2021 | ||
da184284 AV |
2022 | static int copy_msghdr_from_user(struct msghdr *kmsg, |
2023 | struct user_msghdr __user *umsg, | |
2024 | struct sockaddr __user **save_addr, | |
2025 | struct iovec **iov) | |
1661bf36 | 2026 | { |
ffb07550 | 2027 | struct user_msghdr msg; |
08adb7da AV |
2028 | ssize_t err; |
2029 | ||
ffb07550 | 2030 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 2031 | return -EFAULT; |
dbb490b9 | 2032 | |
864d9664 | 2033 | kmsg->msg_control = (void __force *)msg.msg_control; |
ffb07550 AV |
2034 | kmsg->msg_controllen = msg.msg_controllen; |
2035 | kmsg->msg_flags = msg.msg_flags; | |
2036 | ||
2037 | kmsg->msg_namelen = msg.msg_namelen; | |
2038 | if (!msg.msg_name) | |
6a2a2b3a AS |
2039 | kmsg->msg_namelen = 0; |
2040 | ||
dbb490b9 ML |
2041 | if (kmsg->msg_namelen < 0) |
2042 | return -EINVAL; | |
2043 | ||
1661bf36 | 2044 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 2045 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
2046 | |
2047 | if (save_addr) | |
ffb07550 | 2048 | *save_addr = msg.msg_name; |
08adb7da | 2049 | |
ffb07550 | 2050 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 2051 | if (!save_addr) { |
864d9664 PA |
2052 | err = move_addr_to_kernel(msg.msg_name, |
2053 | kmsg->msg_namelen, | |
08adb7da AV |
2054 | kmsg->msg_name); |
2055 | if (err < 0) | |
2056 | return err; | |
2057 | } | |
2058 | } else { | |
2059 | kmsg->msg_name = NULL; | |
2060 | kmsg->msg_namelen = 0; | |
2061 | } | |
2062 | ||
ffb07550 | 2063 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
2064 | return -EMSGSIZE; |
2065 | ||
0345f931 | 2066 | kmsg->msg_iocb = NULL; |
2067 | ||
ffb07550 AV |
2068 | return import_iovec(save_addr ? READ : WRITE, |
2069 | msg.msg_iov, msg.msg_iovlen, | |
da184284 | 2070 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
1661bf36 DC |
2071 | } |
2072 | ||
666547ff | 2073 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2074 | struct msghdr *msg_sys, unsigned int flags, |
28a94d8f TH |
2075 | struct used_address *used_address, |
2076 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 2077 | { |
89bddce5 SH |
2078 | struct compat_msghdr __user *msg_compat = |
2079 | (struct compat_msghdr __user *)msg; | |
230b1839 | 2080 | struct sockaddr_storage address; |
1da177e4 | 2081 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 2082 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 2083 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 2084 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 2085 | unsigned char *ctl_buf = ctl; |
d8725c86 | 2086 | int ctl_len; |
08adb7da | 2087 | ssize_t err; |
89bddce5 | 2088 | |
08adb7da | 2089 | msg_sys->msg_name = &address; |
1da177e4 | 2090 | |
08449320 | 2091 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2092 | err = get_compat_msghdr(msg_sys, msg_compat, NULL, &iov); |
08449320 | 2093 | else |
08adb7da | 2094 | err = copy_msghdr_from_user(msg_sys, msg, NULL, &iov); |
89bddce5 | 2095 | if (err < 0) |
da184284 | 2096 | return err; |
1da177e4 LT |
2097 | |
2098 | err = -ENOBUFS; | |
2099 | ||
228e548e | 2100 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 2101 | goto out_freeiov; |
28a94d8f | 2102 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2103 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2104 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2105 | err = |
228e548e | 2106 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2107 | sizeof(ctl)); |
1da177e4 LT |
2108 | if (err) |
2109 | goto out_freeiov; | |
228e548e AB |
2110 | ctl_buf = msg_sys->msg_control; |
2111 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2112 | } else if (ctl_len) { |
ac4340fc DM |
2113 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2114 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2115 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2116 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2117 | if (ctl_buf == NULL) |
1da177e4 LT |
2118 | goto out_freeiov; |
2119 | } | |
2120 | err = -EFAULT; | |
2121 | /* | |
228e548e | 2122 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2123 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2124 | * checking falls down on this. | |
2125 | */ | |
fb8621bb | 2126 | if (copy_from_user(ctl_buf, |
228e548e | 2127 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2128 | ctl_len)) |
1da177e4 | 2129 | goto out_freectl; |
228e548e | 2130 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2131 | } |
228e548e | 2132 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2133 | |
2134 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2135 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2136 | /* |
2137 | * If this is sendmmsg() and current destination address is same as | |
2138 | * previously succeeded address, omit asking LSM's decision. | |
2139 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2140 | * destination address never matches. | |
2141 | */ | |
bc909d9d MD |
2142 | if (used_address && msg_sys->msg_name && |
2143 | used_address->name_len == msg_sys->msg_namelen && | |
2144 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2145 | used_address->name_len)) { |
d8725c86 | 2146 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2147 | goto out_freectl; |
2148 | } | |
d8725c86 | 2149 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2150 | /* |
2151 | * If this is sendmmsg() and sending to current destination address was | |
2152 | * successful, remember it. | |
2153 | */ | |
2154 | if (used_address && err >= 0) { | |
2155 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2156 | if (msg_sys->msg_name) |
2157 | memcpy(&used_address->name, msg_sys->msg_name, | |
2158 | used_address->name_len); | |
c71d8ebe | 2159 | } |
1da177e4 LT |
2160 | |
2161 | out_freectl: | |
89bddce5 | 2162 | if (ctl_buf != ctl) |
1da177e4 LT |
2163 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
2164 | out_freeiov: | |
da184284 | 2165 | kfree(iov); |
228e548e AB |
2166 | return err; |
2167 | } | |
2168 | ||
2169 | /* | |
2170 | * BSD sendmsg interface | |
2171 | */ | |
2172 | ||
e1834a32 DB |
2173 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2174 | bool forbid_cmsg_compat) | |
228e548e AB |
2175 | { |
2176 | int fput_needed, err; | |
2177 | struct msghdr msg_sys; | |
1be374a0 AL |
2178 | struct socket *sock; |
2179 | ||
e1834a32 DB |
2180 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2181 | return -EINVAL; | |
2182 | ||
1be374a0 | 2183 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2184 | if (!sock) |
2185 | goto out; | |
2186 | ||
28a94d8f | 2187 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2188 | |
6cb153ca | 2189 | fput_light(sock->file, fput_needed); |
89bddce5 | 2190 | out: |
1da177e4 LT |
2191 | return err; |
2192 | } | |
2193 | ||
666547ff | 2194 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 | 2195 | { |
e1834a32 | 2196 | return __sys_sendmsg(fd, msg, flags, true); |
a7526eb5 AL |
2197 | } |
2198 | ||
228e548e AB |
2199 | /* |
2200 | * Linux sendmmsg interface | |
2201 | */ | |
2202 | ||
2203 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
e1834a32 | 2204 | unsigned int flags, bool forbid_cmsg_compat) |
228e548e AB |
2205 | { |
2206 | int fput_needed, err, datagrams; | |
2207 | struct socket *sock; | |
2208 | struct mmsghdr __user *entry; | |
2209 | struct compat_mmsghdr __user *compat_entry; | |
2210 | struct msghdr msg_sys; | |
c71d8ebe | 2211 | struct used_address used_address; |
f092276d | 2212 | unsigned int oflags = flags; |
228e548e | 2213 | |
e1834a32 DB |
2214 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2215 | return -EINVAL; | |
2216 | ||
98382f41 AB |
2217 | if (vlen > UIO_MAXIOV) |
2218 | vlen = UIO_MAXIOV; | |
228e548e AB |
2219 | |
2220 | datagrams = 0; | |
2221 | ||
2222 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2223 | if (!sock) | |
2224 | return err; | |
2225 | ||
c71d8ebe | 2226 | used_address.name_len = UINT_MAX; |
228e548e AB |
2227 | entry = mmsg; |
2228 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2229 | err = 0; |
f092276d | 2230 | flags |= MSG_BATCH; |
228e548e AB |
2231 | |
2232 | while (datagrams < vlen) { | |
f092276d TH |
2233 | if (datagrams == vlen - 1) |
2234 | flags = oflags; | |
2235 | ||
228e548e | 2236 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2237 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2238 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2239 | if (err < 0) |
2240 | break; | |
2241 | err = __put_user(err, &compat_entry->msg_len); | |
2242 | ++compat_entry; | |
2243 | } else { | |
a7526eb5 | 2244 | err = ___sys_sendmsg(sock, |
666547ff | 2245 | (struct user_msghdr __user *)entry, |
28a94d8f | 2246 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2247 | if (err < 0) |
2248 | break; | |
2249 | err = put_user(err, &entry->msg_len); | |
2250 | ++entry; | |
2251 | } | |
2252 | ||
2253 | if (err) | |
2254 | break; | |
2255 | ++datagrams; | |
3023898b SHY |
2256 | if (msg_data_left(&msg_sys)) |
2257 | break; | |
a78cb84c | 2258 | cond_resched(); |
228e548e AB |
2259 | } |
2260 | ||
228e548e AB |
2261 | fput_light(sock->file, fput_needed); |
2262 | ||
728ffb86 AB |
2263 | /* We only return an error if no datagrams were able to be sent */ |
2264 | if (datagrams != 0) | |
228e548e AB |
2265 | return datagrams; |
2266 | ||
228e548e AB |
2267 | return err; |
2268 | } | |
2269 | ||
2270 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2271 | unsigned int, vlen, unsigned int, flags) | |
2272 | { | |
e1834a32 | 2273 | return __sys_sendmmsg(fd, mmsg, vlen, flags, true); |
228e548e AB |
2274 | } |
2275 | ||
666547ff | 2276 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2277 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2278 | { |
89bddce5 SH |
2279 | struct compat_msghdr __user *msg_compat = |
2280 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2281 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2282 | struct iovec *iov = iovstack; |
1da177e4 | 2283 | unsigned long cmsg_ptr; |
2da62906 | 2284 | int len; |
08adb7da | 2285 | ssize_t err; |
1da177e4 LT |
2286 | |
2287 | /* kernel mode address */ | |
230b1839 | 2288 | struct sockaddr_storage addr; |
1da177e4 LT |
2289 | |
2290 | /* user mode address pointers */ | |
2291 | struct sockaddr __user *uaddr; | |
08adb7da | 2292 | int __user *uaddr_len = COMPAT_NAMELEN(msg); |
89bddce5 | 2293 | |
08adb7da | 2294 | msg_sys->msg_name = &addr; |
1da177e4 | 2295 | |
f3d33426 | 2296 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2297 | err = get_compat_msghdr(msg_sys, msg_compat, &uaddr, &iov); |
f3d33426 | 2298 | else |
08adb7da | 2299 | err = copy_msghdr_from_user(msg_sys, msg, &uaddr, &iov); |
1da177e4 | 2300 | if (err < 0) |
da184284 | 2301 | return err; |
1da177e4 | 2302 | |
a2e27255 ACM |
2303 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2304 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2305 | |
f3d33426 HFS |
2306 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2307 | msg_sys->msg_namelen = 0; | |
2308 | ||
1da177e4 LT |
2309 | if (sock->file->f_flags & O_NONBLOCK) |
2310 | flags |= MSG_DONTWAIT; | |
2da62906 | 2311 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, flags); |
1da177e4 LT |
2312 | if (err < 0) |
2313 | goto out_freeiov; | |
2314 | len = err; | |
2315 | ||
2316 | if (uaddr != NULL) { | |
43db362d | 2317 | err = move_addr_to_user(&addr, |
a2e27255 | 2318 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2319 | uaddr_len); |
1da177e4 LT |
2320 | if (err < 0) |
2321 | goto out_freeiov; | |
2322 | } | |
a2e27255 | 2323 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2324 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2325 | if (err) |
2326 | goto out_freeiov; | |
2327 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2328 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2329 | &msg_compat->msg_controllen); |
2330 | else | |
a2e27255 | 2331 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2332 | &msg->msg_controllen); |
2333 | if (err) | |
2334 | goto out_freeiov; | |
2335 | err = len; | |
2336 | ||
2337 | out_freeiov: | |
da184284 | 2338 | kfree(iov); |
a2e27255 ACM |
2339 | return err; |
2340 | } | |
2341 | ||
2342 | /* | |
2343 | * BSD recvmsg interface | |
2344 | */ | |
2345 | ||
e1834a32 DB |
2346 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2347 | bool forbid_cmsg_compat) | |
a2e27255 ACM |
2348 | { |
2349 | int fput_needed, err; | |
2350 | struct msghdr msg_sys; | |
1be374a0 AL |
2351 | struct socket *sock; |
2352 | ||
e1834a32 DB |
2353 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2354 | return -EINVAL; | |
2355 | ||
1be374a0 | 2356 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2357 | if (!sock) |
2358 | goto out; | |
2359 | ||
a7526eb5 | 2360 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2361 | |
6cb153ca | 2362 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2363 | out: |
2364 | return err; | |
2365 | } | |
2366 | ||
666547ff | 2367 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2368 | unsigned int, flags) |
2369 | { | |
e1834a32 | 2370 | return __sys_recvmsg(fd, msg, flags, true); |
a7526eb5 AL |
2371 | } |
2372 | ||
a2e27255 ACM |
2373 | /* |
2374 | * Linux recvmmsg interface | |
2375 | */ | |
2376 | ||
2377 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2378 | unsigned int flags, struct timespec *timeout) | |
2379 | { | |
2380 | int fput_needed, err, datagrams; | |
2381 | struct socket *sock; | |
2382 | struct mmsghdr __user *entry; | |
d7256d0e | 2383 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2384 | struct msghdr msg_sys; |
766b9f92 DD |
2385 | struct timespec64 end_time; |
2386 | struct timespec64 timeout64; | |
a2e27255 ACM |
2387 | |
2388 | if (timeout && | |
2389 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2390 | timeout->tv_nsec)) | |
2391 | return -EINVAL; | |
2392 | ||
2393 | datagrams = 0; | |
2394 | ||
2395 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2396 | if (!sock) | |
2397 | return err; | |
2398 | ||
7797dc41 SHY |
2399 | if (likely(!(flags & MSG_ERRQUEUE))) { |
2400 | err = sock_error(sock->sk); | |
2401 | if (err) { | |
2402 | datagrams = err; | |
2403 | goto out_put; | |
2404 | } | |
e623a9e9 | 2405 | } |
a2e27255 ACM |
2406 | |
2407 | entry = mmsg; | |
d7256d0e | 2408 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2409 | |
2410 | while (datagrams < vlen) { | |
2411 | /* | |
2412 | * No need to ask LSM for more than the first datagram. | |
2413 | */ | |
d7256d0e | 2414 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2415 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2416 | &msg_sys, flags & ~MSG_WAITFORONE, |
2417 | datagrams); | |
d7256d0e JMG |
2418 | if (err < 0) |
2419 | break; | |
2420 | err = __put_user(err, &compat_entry->msg_len); | |
2421 | ++compat_entry; | |
2422 | } else { | |
a7526eb5 | 2423 | err = ___sys_recvmsg(sock, |
666547ff | 2424 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2425 | &msg_sys, flags & ~MSG_WAITFORONE, |
2426 | datagrams); | |
d7256d0e JMG |
2427 | if (err < 0) |
2428 | break; | |
2429 | err = put_user(err, &entry->msg_len); | |
2430 | ++entry; | |
2431 | } | |
2432 | ||
a2e27255 ACM |
2433 | if (err) |
2434 | break; | |
a2e27255 ACM |
2435 | ++datagrams; |
2436 | ||
71c5c159 BB |
2437 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2438 | if (flags & MSG_WAITFORONE) | |
2439 | flags |= MSG_DONTWAIT; | |
2440 | ||
a2e27255 | 2441 | if (timeout) { |
766b9f92 DD |
2442 | ktime_get_ts64(&timeout64); |
2443 | *timeout = timespec64_to_timespec( | |
2444 | timespec64_sub(end_time, timeout64)); | |
a2e27255 ACM |
2445 | if (timeout->tv_sec < 0) { |
2446 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2447 | break; | |
2448 | } | |
2449 | ||
2450 | /* Timeout, return less than vlen datagrams */ | |
2451 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2452 | break; | |
2453 | } | |
2454 | ||
2455 | /* Out of band data, return right away */ | |
2456 | if (msg_sys.msg_flags & MSG_OOB) | |
2457 | break; | |
a78cb84c | 2458 | cond_resched(); |
a2e27255 ACM |
2459 | } |
2460 | ||
a2e27255 | 2461 | if (err == 0) |
34b88a68 ACM |
2462 | goto out_put; |
2463 | ||
2464 | if (datagrams == 0) { | |
2465 | datagrams = err; | |
2466 | goto out_put; | |
2467 | } | |
a2e27255 | 2468 | |
34b88a68 ACM |
2469 | /* |
2470 | * We may return less entries than requested (vlen) if the | |
2471 | * sock is non block and there aren't enough datagrams... | |
2472 | */ | |
2473 | if (err != -EAGAIN) { | |
a2e27255 | 2474 | /* |
34b88a68 ACM |
2475 | * ... or if recvmsg returns an error after we |
2476 | * received some datagrams, where we record the | |
2477 | * error to return on the next call or if the | |
2478 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2479 | */ |
34b88a68 | 2480 | sock->sk->sk_err = -err; |
a2e27255 | 2481 | } |
34b88a68 ACM |
2482 | out_put: |
2483 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2484 | |
34b88a68 | 2485 | return datagrams; |
a2e27255 ACM |
2486 | } |
2487 | ||
1255e269 DB |
2488 | static int do_sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2489 | unsigned int vlen, unsigned int flags, | |
2490 | struct timespec __user *timeout) | |
a2e27255 ACM |
2491 | { |
2492 | int datagrams; | |
2493 | struct timespec timeout_sys; | |
2494 | ||
1be374a0 AL |
2495 | if (flags & MSG_CMSG_COMPAT) |
2496 | return -EINVAL; | |
2497 | ||
a2e27255 ACM |
2498 | if (!timeout) |
2499 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2500 | ||
2501 | if (copy_from_user(&timeout_sys, timeout, sizeof(timeout_sys))) | |
2502 | return -EFAULT; | |
2503 | ||
2504 | datagrams = __sys_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
2505 | ||
2506 | if (datagrams > 0 && | |
2507 | copy_to_user(timeout, &timeout_sys, sizeof(timeout_sys))) | |
2508 | datagrams = -EFAULT; | |
2509 | ||
2510 | return datagrams; | |
2511 | } | |
2512 | ||
1255e269 DB |
2513 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, |
2514 | unsigned int, vlen, unsigned int, flags, | |
2515 | struct timespec __user *, timeout) | |
2516 | { | |
2517 | return do_sys_recvmmsg(fd, mmsg, vlen, flags, timeout); | |
2518 | } | |
2519 | ||
a2e27255 | 2520 | #ifdef __ARCH_WANT_SYS_SOCKETCALL |
1da177e4 LT |
2521 | /* Argument list sizes for sys_socketcall */ |
2522 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2523 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2524 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2525 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2526 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2527 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2528 | }; |
2529 | ||
1da177e4 LT |
2530 | #undef AL |
2531 | ||
2532 | /* | |
89bddce5 | 2533 | * System call vectors. |
1da177e4 LT |
2534 | * |
2535 | * Argument checking cleaned up. Saved 20% in size. | |
2536 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2537 | * it is set by the callees. |
1da177e4 LT |
2538 | */ |
2539 | ||
3e0fa65f | 2540 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2541 | { |
2950fa9d | 2542 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2543 | unsigned long a0, a1; |
1da177e4 | 2544 | int err; |
47379052 | 2545 | unsigned int len; |
1da177e4 | 2546 | |
228e548e | 2547 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 LT |
2548 | return -EINVAL; |
2549 | ||
47379052 AV |
2550 | len = nargs[call]; |
2551 | if (len > sizeof(a)) | |
2552 | return -EINVAL; | |
2553 | ||
1da177e4 | 2554 | /* copy_from_user should be SMP safe. */ |
47379052 | 2555 | if (copy_from_user(a, args, len)) |
1da177e4 | 2556 | return -EFAULT; |
3ec3b2fb | 2557 | |
2950fa9d CG |
2558 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2559 | if (err) | |
2560 | return err; | |
3ec3b2fb | 2561 | |
89bddce5 SH |
2562 | a0 = a[0]; |
2563 | a1 = a[1]; | |
2564 | ||
2565 | switch (call) { | |
2566 | case SYS_SOCKET: | |
9d6a15c3 | 2567 | err = __sys_socket(a0, a1, a[2]); |
89bddce5 SH |
2568 | break; |
2569 | case SYS_BIND: | |
a87d35d8 | 2570 | err = __sys_bind(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2571 | break; |
2572 | case SYS_CONNECT: | |
1387c2c2 | 2573 | err = __sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2574 | break; |
2575 | case SYS_LISTEN: | |
25e290ee | 2576 | err = __sys_listen(a0, a1); |
89bddce5 SH |
2577 | break; |
2578 | case SYS_ACCEPT: | |
4541e805 DB |
2579 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2580 | (int __user *)a[2], 0); | |
89bddce5 SH |
2581 | break; |
2582 | case SYS_GETSOCKNAME: | |
2583 | err = | |
8882a107 DB |
2584 | __sys_getsockname(a0, (struct sockaddr __user *)a1, |
2585 | (int __user *)a[2]); | |
89bddce5 SH |
2586 | break; |
2587 | case SYS_GETPEERNAME: | |
2588 | err = | |
b21c8f83 DB |
2589 | __sys_getpeername(a0, (struct sockaddr __user *)a1, |
2590 | (int __user *)a[2]); | |
89bddce5 SH |
2591 | break; |
2592 | case SYS_SOCKETPAIR: | |
6debc8d8 | 2593 | err = __sys_socketpair(a0, a1, a[2], (int __user *)a[3]); |
89bddce5 SH |
2594 | break; |
2595 | case SYS_SEND: | |
f3bf896b DB |
2596 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2597 | NULL, 0); | |
89bddce5 SH |
2598 | break; |
2599 | case SYS_SENDTO: | |
211b634b DB |
2600 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2601 | (struct sockaddr __user *)a[4], a[5]); | |
89bddce5 SH |
2602 | break; |
2603 | case SYS_RECV: | |
d27e9afc DB |
2604 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2605 | NULL, NULL); | |
89bddce5 SH |
2606 | break; |
2607 | case SYS_RECVFROM: | |
7a09e1eb DB |
2608 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2609 | (struct sockaddr __user *)a[4], | |
2610 | (int __user *)a[5]); | |
89bddce5 SH |
2611 | break; |
2612 | case SYS_SHUTDOWN: | |
005a1aea | 2613 | err = __sys_shutdown(a0, a1); |
89bddce5 SH |
2614 | break; |
2615 | case SYS_SETSOCKOPT: | |
cc36dca0 DB |
2616 | err = __sys_setsockopt(a0, a1, a[2], (char __user *)a[3], |
2617 | a[4]); | |
89bddce5 SH |
2618 | break; |
2619 | case SYS_GETSOCKOPT: | |
2620 | err = | |
13a2d70e DB |
2621 | __sys_getsockopt(a0, a1, a[2], (char __user *)a[3], |
2622 | (int __user *)a[4]); | |
89bddce5 SH |
2623 | break; |
2624 | case SYS_SENDMSG: | |
e1834a32 DB |
2625 | err = __sys_sendmsg(a0, (struct user_msghdr __user *)a1, |
2626 | a[2], true); | |
89bddce5 | 2627 | break; |
228e548e | 2628 | case SYS_SENDMMSG: |
e1834a32 DB |
2629 | err = __sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
2630 | a[3], true); | |
228e548e | 2631 | break; |
89bddce5 | 2632 | case SYS_RECVMSG: |
e1834a32 DB |
2633 | err = __sys_recvmsg(a0, (struct user_msghdr __user *)a1, |
2634 | a[2], true); | |
89bddce5 | 2635 | break; |
a2e27255 | 2636 | case SYS_RECVMMSG: |
1255e269 DB |
2637 | err = do_sys_recvmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
2638 | a[3], (struct timespec __user *)a[4]); | |
a2e27255 | 2639 | break; |
de11defe | 2640 | case SYS_ACCEPT4: |
4541e805 DB |
2641 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2642 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2643 | break; |
89bddce5 SH |
2644 | default: |
2645 | err = -EINVAL; | |
2646 | break; | |
1da177e4 LT |
2647 | } |
2648 | return err; | |
2649 | } | |
2650 | ||
89bddce5 | 2651 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2652 | |
55737fda SH |
2653 | /** |
2654 | * sock_register - add a socket protocol handler | |
2655 | * @ops: description of protocol | |
2656 | * | |
1da177e4 LT |
2657 | * This function is called by a protocol handler that wants to |
2658 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2659 | * socket interface. The value ops->family corresponds to the |
55737fda | 2660 | * socket system call protocol family. |
1da177e4 | 2661 | */ |
f0fd27d4 | 2662 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2663 | { |
2664 | int err; | |
2665 | ||
2666 | if (ops->family >= NPROTO) { | |
3410f22e | 2667 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2668 | return -ENOBUFS; |
2669 | } | |
55737fda SH |
2670 | |
2671 | spin_lock(&net_family_lock); | |
190683a9 ED |
2672 | if (rcu_dereference_protected(net_families[ops->family], |
2673 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2674 | err = -EEXIST; |
2675 | else { | |
cf778b00 | 2676 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2677 | err = 0; |
2678 | } | |
55737fda SH |
2679 | spin_unlock(&net_family_lock); |
2680 | ||
3410f22e | 2681 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2682 | return err; |
2683 | } | |
c6d409cf | 2684 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2685 | |
55737fda SH |
2686 | /** |
2687 | * sock_unregister - remove a protocol handler | |
2688 | * @family: protocol family to remove | |
2689 | * | |
1da177e4 LT |
2690 | * This function is called by a protocol handler that wants to |
2691 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2692 | * new socket creation. |
2693 | * | |
2694 | * If protocol handler is a module, then it can use module reference | |
2695 | * counts to protect against new references. If protocol handler is not | |
2696 | * a module then it needs to provide its own protection in | |
2697 | * the ops->create routine. | |
1da177e4 | 2698 | */ |
f0fd27d4 | 2699 | void sock_unregister(int family) |
1da177e4 | 2700 | { |
f0fd27d4 | 2701 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2702 | |
55737fda | 2703 | spin_lock(&net_family_lock); |
a9b3cd7f | 2704 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2705 | spin_unlock(&net_family_lock); |
2706 | ||
2707 | synchronize_rcu(); | |
2708 | ||
3410f22e | 2709 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2710 | } |
c6d409cf | 2711 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2712 | |
bf2ae2e4 XL |
2713 | bool sock_is_registered(int family) |
2714 | { | |
2715 | return family < NPROTO && rcu_access_pointer(net_families[family]); | |
2716 | } | |
2717 | ||
77d76ea3 | 2718 | static int __init sock_init(void) |
1da177e4 | 2719 | { |
b3e19d92 | 2720 | int err; |
2ca794e5 EB |
2721 | /* |
2722 | * Initialize the network sysctl infrastructure. | |
2723 | */ | |
2724 | err = net_sysctl_init(); | |
2725 | if (err) | |
2726 | goto out; | |
b3e19d92 | 2727 | |
1da177e4 | 2728 | /* |
89bddce5 | 2729 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2730 | */ |
2731 | skb_init(); | |
1da177e4 LT |
2732 | |
2733 | /* | |
89bddce5 | 2734 | * Initialize the protocols module. |
1da177e4 LT |
2735 | */ |
2736 | ||
2737 | init_inodecache(); | |
b3e19d92 NP |
2738 | |
2739 | err = register_filesystem(&sock_fs_type); | |
2740 | if (err) | |
2741 | goto out_fs; | |
1da177e4 | 2742 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2743 | if (IS_ERR(sock_mnt)) { |
2744 | err = PTR_ERR(sock_mnt); | |
2745 | goto out_mount; | |
2746 | } | |
77d76ea3 AK |
2747 | |
2748 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2749 | */ |
2750 | ||
2751 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
2752 | err = netfilter_init(); |
2753 | if (err) | |
2754 | goto out; | |
1da177e4 | 2755 | #endif |
cbeb321a | 2756 | |
408eccce | 2757 | ptp_classifier_init(); |
c1f19b51 | 2758 | |
b3e19d92 NP |
2759 | out: |
2760 | return err; | |
2761 | ||
2762 | out_mount: | |
2763 | unregister_filesystem(&sock_fs_type); | |
2764 | out_fs: | |
2765 | goto out; | |
1da177e4 LT |
2766 | } |
2767 | ||
77d76ea3 AK |
2768 | core_initcall(sock_init); /* early initcall */ |
2769 | ||
1da177e4 LT |
2770 | #ifdef CONFIG_PROC_FS |
2771 | void socket_seq_show(struct seq_file *seq) | |
2772 | { | |
648845ab TZ |
2773 | seq_printf(seq, "sockets: used %d\n", |
2774 | sock_inuse_get(seq->private)); | |
1da177e4 | 2775 | } |
89bddce5 | 2776 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 2777 | |
89bbfc95 | 2778 | #ifdef CONFIG_COMPAT |
6b96018b | 2779 | static int do_siocgstamp(struct net *net, struct socket *sock, |
644595f8 | 2780 | unsigned int cmd, void __user *up) |
7a229387 | 2781 | { |
7a229387 AB |
2782 | mm_segment_t old_fs = get_fs(); |
2783 | struct timeval ktv; | |
2784 | int err; | |
2785 | ||
2786 | set_fs(KERNEL_DS); | |
6b96018b | 2787 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&ktv); |
7a229387 | 2788 | set_fs(old_fs); |
644595f8 | 2789 | if (!err) |
ed6fe9d6 | 2790 | err = compat_put_timeval(&ktv, up); |
644595f8 | 2791 | |
7a229387 AB |
2792 | return err; |
2793 | } | |
2794 | ||
6b96018b | 2795 | static int do_siocgstampns(struct net *net, struct socket *sock, |
644595f8 | 2796 | unsigned int cmd, void __user *up) |
7a229387 | 2797 | { |
7a229387 AB |
2798 | mm_segment_t old_fs = get_fs(); |
2799 | struct timespec kts; | |
2800 | int err; | |
2801 | ||
2802 | set_fs(KERNEL_DS); | |
6b96018b | 2803 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&kts); |
7a229387 | 2804 | set_fs(old_fs); |
644595f8 | 2805 | if (!err) |
ed6fe9d6 | 2806 | err = compat_put_timespec(&kts, up); |
644595f8 | 2807 | |
7a229387 AB |
2808 | return err; |
2809 | } | |
2810 | ||
36fd633e | 2811 | static int compat_dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 2812 | { |
6b96018b | 2813 | struct compat_ifconf ifc32; |
7a229387 | 2814 | struct ifconf ifc; |
7a229387 AB |
2815 | int err; |
2816 | ||
6b96018b | 2817 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2818 | return -EFAULT; |
2819 | ||
36fd633e AV |
2820 | ifc.ifc_len = ifc32.ifc_len; |
2821 | ifc.ifc_req = compat_ptr(ifc32.ifcbuf); | |
7a229387 | 2822 | |
36fd633e AV |
2823 | rtnl_lock(); |
2824 | err = dev_ifconf(net, &ifc, sizeof(struct compat_ifreq)); | |
2825 | rtnl_unlock(); | |
7a229387 AB |
2826 | if (err) |
2827 | return err; | |
2828 | ||
36fd633e | 2829 | ifc32.ifc_len = ifc.ifc_len; |
6b96018b | 2830 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2831 | return -EFAULT; |
2832 | ||
2833 | return 0; | |
2834 | } | |
2835 | ||
6b96018b | 2836 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 2837 | { |
3a7da39d BH |
2838 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
2839 | bool convert_in = false, convert_out = false; | |
44c02a2c AV |
2840 | size_t buf_size = 0; |
2841 | struct ethtool_rxnfc __user *rxnfc = NULL; | |
2842 | struct ifreq ifr; | |
3a7da39d BH |
2843 | u32 rule_cnt = 0, actual_rule_cnt; |
2844 | u32 ethcmd; | |
7a229387 | 2845 | u32 data; |
3a7da39d | 2846 | int ret; |
7a229387 | 2847 | |
3a7da39d BH |
2848 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
2849 | return -EFAULT; | |
7a229387 | 2850 | |
3a7da39d BH |
2851 | compat_rxnfc = compat_ptr(data); |
2852 | ||
2853 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
2854 | return -EFAULT; |
2855 | ||
3a7da39d BH |
2856 | /* Most ethtool structures are defined without padding. |
2857 | * Unfortunately struct ethtool_rxnfc is an exception. | |
2858 | */ | |
2859 | switch (ethcmd) { | |
2860 | default: | |
2861 | break; | |
2862 | case ETHTOOL_GRXCLSRLALL: | |
2863 | /* Buffer size is variable */ | |
2864 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
2865 | return -EFAULT; | |
2866 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
2867 | return -ENOMEM; | |
2868 | buf_size += rule_cnt * sizeof(u32); | |
2869 | /* fall through */ | |
2870 | case ETHTOOL_GRXRINGS: | |
2871 | case ETHTOOL_GRXCLSRLCNT: | |
2872 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 2873 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
2874 | convert_out = true; |
2875 | /* fall through */ | |
2876 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
2877 | buf_size += sizeof(struct ethtool_rxnfc); |
2878 | convert_in = true; | |
44c02a2c | 2879 | rxnfc = compat_alloc_user_space(buf_size); |
3a7da39d BH |
2880 | break; |
2881 | } | |
2882 | ||
44c02a2c | 2883 | if (copy_from_user(&ifr.ifr_name, &ifr32->ifr_name, IFNAMSIZ)) |
7a229387 AB |
2884 | return -EFAULT; |
2885 | ||
44c02a2c | 2886 | ifr.ifr_data = convert_in ? rxnfc : (void __user *)compat_rxnfc; |
7a229387 | 2887 | |
3a7da39d | 2888 | if (convert_in) { |
127fe533 | 2889 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
2890 | * fs.ring_cookie and at the end of fs, but nowhere else. |
2891 | */ | |
127fe533 AD |
2892 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
2893 | sizeof(compat_rxnfc->fs.m_ext) != | |
2894 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
2895 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
2896 | BUILD_BUG_ON( |
2897 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
2898 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
2899 | offsetof(struct ethtool_rxnfc, fs.location) - | |
2900 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
2901 | ||
2902 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
2903 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
2904 | (void __user *)rxnfc) || | |
3a7da39d BH |
2905 | copy_in_user(&rxnfc->fs.ring_cookie, |
2906 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 SH |
2907 | (void __user *)(&rxnfc->fs.location + 1) - |
2908 | (void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2909 | copy_in_user(&rxnfc->rule_cnt, &compat_rxnfc->rule_cnt, |
2910 | sizeof(rxnfc->rule_cnt))) | |
2911 | return -EFAULT; | |
2912 | } | |
2913 | ||
44c02a2c | 2914 | ret = dev_ioctl(net, SIOCETHTOOL, &ifr, NULL); |
3a7da39d BH |
2915 | if (ret) |
2916 | return ret; | |
2917 | ||
2918 | if (convert_out) { | |
2919 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
2920 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
2921 | (const void __user *)rxnfc) || | |
3a7da39d BH |
2922 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
2923 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
2924 | (const void __user *)(&rxnfc->fs.location + 1) - |
2925 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2926 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
2927 | sizeof(rxnfc->rule_cnt))) | |
2928 | return -EFAULT; | |
2929 | ||
2930 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
2931 | /* As an optimisation, we only copy the actual | |
2932 | * number of rules that the underlying | |
2933 | * function returned. Since Mallory might | |
2934 | * change the rule count in user memory, we | |
2935 | * check that it is less than the rule count | |
2936 | * originally given (as the user buffer size), | |
2937 | * which has been range-checked. | |
2938 | */ | |
2939 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
2940 | return -EFAULT; | |
2941 | if (actual_rule_cnt < rule_cnt) | |
2942 | rule_cnt = actual_rule_cnt; | |
2943 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
2944 | &rxnfc->rule_locs[0], | |
2945 | rule_cnt * sizeof(u32))) | |
2946 | return -EFAULT; | |
2947 | } | |
2948 | } | |
2949 | ||
2950 | return 0; | |
7a229387 AB |
2951 | } |
2952 | ||
7a50a240 AB |
2953 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
2954 | { | |
7a50a240 | 2955 | compat_uptr_t uptr32; |
44c02a2c AV |
2956 | struct ifreq ifr; |
2957 | void __user *saved; | |
2958 | int err; | |
7a50a240 | 2959 | |
44c02a2c | 2960 | if (copy_from_user(&ifr, uifr32, sizeof(struct compat_ifreq))) |
7a50a240 AB |
2961 | return -EFAULT; |
2962 | ||
2963 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
2964 | return -EFAULT; | |
2965 | ||
44c02a2c AV |
2966 | saved = ifr.ifr_settings.ifs_ifsu.raw_hdlc; |
2967 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = compat_ptr(uptr32); | |
7a229387 | 2968 | |
44c02a2c AV |
2969 | err = dev_ioctl(net, SIOCWANDEV, &ifr, NULL); |
2970 | if (!err) { | |
2971 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = saved; | |
2972 | if (copy_to_user(uifr32, &ifr, sizeof(struct compat_ifreq))) | |
2973 | err = -EFAULT; | |
ccbd6a5a | 2974 | } |
44c02a2c | 2975 | return err; |
7a229387 AB |
2976 | } |
2977 | ||
590d4693 BH |
2978 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
2979 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 2980 | struct compat_ifreq __user *u_ifreq32) |
7a229387 | 2981 | { |
44c02a2c | 2982 | struct ifreq ifreq; |
7a229387 AB |
2983 | u32 data32; |
2984 | ||
44c02a2c | 2985 | if (copy_from_user(ifreq.ifr_name, u_ifreq32->ifr_name, IFNAMSIZ)) |
7a229387 | 2986 | return -EFAULT; |
44c02a2c | 2987 | if (get_user(data32, &u_ifreq32->ifr_data)) |
7a229387 | 2988 | return -EFAULT; |
44c02a2c | 2989 | ifreq.ifr_data = compat_ptr(data32); |
7a229387 | 2990 | |
44c02a2c | 2991 | return dev_ioctl(net, cmd, &ifreq, NULL); |
7a229387 AB |
2992 | } |
2993 | ||
a2116ed2 AB |
2994 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
2995 | struct compat_ifreq __user *uifr32) | |
2996 | { | |
2997 | struct ifreq ifr; | |
2998 | struct compat_ifmap __user *uifmap32; | |
a2116ed2 AB |
2999 | int err; |
3000 | ||
3001 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
3002 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3003 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3004 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3005 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3006 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3007 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3008 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3009 | if (err) |
3010 | return -EFAULT; | |
3011 | ||
44c02a2c | 3012 | err = dev_ioctl(net, cmd, &ifr, NULL); |
a2116ed2 AB |
3013 | |
3014 | if (cmd == SIOCGIFMAP && !err) { | |
3015 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3016 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3017 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3018 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3019 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3020 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3021 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3022 | if (err) |
3023 | err = -EFAULT; | |
3024 | } | |
3025 | return err; | |
3026 | } | |
3027 | ||
7a229387 | 3028 | struct rtentry32 { |
c6d409cf | 3029 | u32 rt_pad1; |
7a229387 AB |
3030 | struct sockaddr rt_dst; /* target address */ |
3031 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
3032 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
3033 | unsigned short rt_flags; |
3034 | short rt_pad2; | |
3035 | u32 rt_pad3; | |
3036 | unsigned char rt_tos; | |
3037 | unsigned char rt_class; | |
3038 | short rt_pad4; | |
3039 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 3040 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
3041 | u32 rt_mtu; /* per route MTU/Window */ |
3042 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
3043 | unsigned short rt_irtt; /* Initial RTT */ |
3044 | }; | |
3045 | ||
3046 | struct in6_rtmsg32 { | |
3047 | struct in6_addr rtmsg_dst; | |
3048 | struct in6_addr rtmsg_src; | |
3049 | struct in6_addr rtmsg_gateway; | |
3050 | u32 rtmsg_type; | |
3051 | u16 rtmsg_dst_len; | |
3052 | u16 rtmsg_src_len; | |
3053 | u32 rtmsg_metric; | |
3054 | u32 rtmsg_info; | |
3055 | u32 rtmsg_flags; | |
3056 | s32 rtmsg_ifindex; | |
3057 | }; | |
3058 | ||
6b96018b AB |
3059 | static int routing_ioctl(struct net *net, struct socket *sock, |
3060 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3061 | { |
3062 | int ret; | |
3063 | void *r = NULL; | |
3064 | struct in6_rtmsg r6; | |
3065 | struct rtentry r4; | |
3066 | char devname[16]; | |
3067 | u32 rtdev; | |
3068 | mm_segment_t old_fs = get_fs(); | |
3069 | ||
6b96018b AB |
3070 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3071 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3072 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3073 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
3074 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3075 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3076 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3077 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3078 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3079 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3080 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3081 | |
3082 | r = (void *) &r6; | |
3083 | } else { /* ipv4 */ | |
6b96018b | 3084 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3085 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3086 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
3087 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
3088 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3089 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3090 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
3091 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3092 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3093 | if (rtdev) { |
c6d409cf | 3094 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3095 | r4.rt_dev = (char __user __force *)devname; |
3096 | devname[15] = 0; | |
7a229387 AB |
3097 | } else |
3098 | r4.rt_dev = NULL; | |
3099 | ||
3100 | r = (void *) &r4; | |
3101 | } | |
3102 | ||
3103 | if (ret) { | |
3104 | ret = -EFAULT; | |
3105 | goto out; | |
3106 | } | |
3107 | ||
c6d409cf | 3108 | set_fs(KERNEL_DS); |
6b96018b | 3109 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3110 | set_fs(old_fs); |
7a229387 AB |
3111 | |
3112 | out: | |
7a229387 AB |
3113 | return ret; |
3114 | } | |
3115 | ||
3116 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3117 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3118 | * use compatible ioctls |
7a229387 | 3119 | */ |
6b96018b | 3120 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3121 | { |
6b96018b | 3122 | compat_ulong_t tmp; |
7a229387 | 3123 | |
6b96018b | 3124 | if (get_user(tmp, argp)) |
7a229387 AB |
3125 | return -EFAULT; |
3126 | if (tmp == BRCTL_GET_VERSION) | |
3127 | return BRCTL_VERSION + 1; | |
3128 | return -EINVAL; | |
3129 | } | |
3130 | ||
6b96018b AB |
3131 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3132 | unsigned int cmd, unsigned long arg) | |
3133 | { | |
3134 | void __user *argp = compat_ptr(arg); | |
3135 | struct sock *sk = sock->sk; | |
3136 | struct net *net = sock_net(sk); | |
7a229387 | 3137 | |
6b96018b | 3138 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3139 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3140 | |
3141 | switch (cmd) { | |
3142 | case SIOCSIFBR: | |
3143 | case SIOCGIFBR: | |
3144 | return old_bridge_ioctl(argp); | |
6b96018b | 3145 | case SIOCGIFCONF: |
36fd633e | 3146 | return compat_dev_ifconf(net, argp); |
6b96018b AB |
3147 | case SIOCETHTOOL: |
3148 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3149 | case SIOCWANDEV: |
3150 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3151 | case SIOCGIFMAP: |
3152 | case SIOCSIFMAP: | |
3153 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3154 | case SIOCADDRT: |
3155 | case SIOCDELRT: | |
3156 | return routing_ioctl(net, sock, cmd, argp); | |
3157 | case SIOCGSTAMP: | |
3158 | return do_siocgstamp(net, sock, cmd, argp); | |
3159 | case SIOCGSTAMPNS: | |
3160 | return do_siocgstampns(net, sock, cmd, argp); | |
590d4693 BH |
3161 | case SIOCBONDSLAVEINFOQUERY: |
3162 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3163 | case SIOCSHWTSTAMP: |
fd468c74 | 3164 | case SIOCGHWTSTAMP: |
590d4693 | 3165 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3166 | |
3167 | case FIOSETOWN: | |
3168 | case SIOCSPGRP: | |
3169 | case FIOGETOWN: | |
3170 | case SIOCGPGRP: | |
3171 | case SIOCBRADDBR: | |
3172 | case SIOCBRDELBR: | |
3173 | case SIOCGIFVLAN: | |
3174 | case SIOCSIFVLAN: | |
3175 | case SIOCADDDLCI: | |
3176 | case SIOCDELDLCI: | |
c62cce2c | 3177 | case SIOCGSKNS: |
6b96018b AB |
3178 | return sock_ioctl(file, cmd, arg); |
3179 | ||
3180 | case SIOCGIFFLAGS: | |
3181 | case SIOCSIFFLAGS: | |
3182 | case SIOCGIFMETRIC: | |
3183 | case SIOCSIFMETRIC: | |
3184 | case SIOCGIFMTU: | |
3185 | case SIOCSIFMTU: | |
3186 | case SIOCGIFMEM: | |
3187 | case SIOCSIFMEM: | |
3188 | case SIOCGIFHWADDR: | |
3189 | case SIOCSIFHWADDR: | |
3190 | case SIOCADDMULTI: | |
3191 | case SIOCDELMULTI: | |
3192 | case SIOCGIFINDEX: | |
6b96018b AB |
3193 | case SIOCGIFADDR: |
3194 | case SIOCSIFADDR: | |
3195 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3196 | case SIOCDIFADDR: |
6b96018b AB |
3197 | case SIOCGIFBRDADDR: |
3198 | case SIOCSIFBRDADDR: | |
3199 | case SIOCGIFDSTADDR: | |
3200 | case SIOCSIFDSTADDR: | |
3201 | case SIOCGIFNETMASK: | |
3202 | case SIOCSIFNETMASK: | |
3203 | case SIOCSIFPFLAGS: | |
3204 | case SIOCGIFPFLAGS: | |
3205 | case SIOCGIFTXQLEN: | |
3206 | case SIOCSIFTXQLEN: | |
3207 | case SIOCBRADDIF: | |
3208 | case SIOCBRDELIF: | |
9177efd3 AB |
3209 | case SIOCSIFNAME: |
3210 | case SIOCGMIIPHY: | |
3211 | case SIOCGMIIREG: | |
3212 | case SIOCSMIIREG: | |
6b96018b AB |
3213 | case SIOCSARP: |
3214 | case SIOCGARP: | |
3215 | case SIOCDARP: | |
6b96018b | 3216 | case SIOCATMARK: |
f92d4fc9 AV |
3217 | case SIOCBONDENSLAVE: |
3218 | case SIOCBONDRELEASE: | |
3219 | case SIOCBONDSETHWADDR: | |
3220 | case SIOCBONDCHANGEACTIVE: | |
4cf808e7 | 3221 | case SIOCGIFNAME: |
9177efd3 AB |
3222 | return sock_do_ioctl(net, sock, cmd, arg); |
3223 | } | |
3224 | ||
6b96018b AB |
3225 | return -ENOIOCTLCMD; |
3226 | } | |
7a229387 | 3227 | |
95c96174 | 3228 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3229 | unsigned long arg) |
89bbfc95 SP |
3230 | { |
3231 | struct socket *sock = file->private_data; | |
3232 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3233 | struct sock *sk; |
3234 | struct net *net; | |
3235 | ||
3236 | sk = sock->sk; | |
3237 | net = sock_net(sk); | |
89bbfc95 SP |
3238 | |
3239 | if (sock->ops->compat_ioctl) | |
3240 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3241 | ||
87de87d5 DM |
3242 | if (ret == -ENOIOCTLCMD && |
3243 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3244 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3245 | ||
6b96018b AB |
3246 | if (ret == -ENOIOCTLCMD) |
3247 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3248 | ||
89bbfc95 SP |
3249 | return ret; |
3250 | } | |
3251 | #endif | |
3252 | ||
ac5a488e SS |
3253 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3254 | { | |
3255 | return sock->ops->bind(sock, addr, addrlen); | |
3256 | } | |
c6d409cf | 3257 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e SS |
3258 | |
3259 | int kernel_listen(struct socket *sock, int backlog) | |
3260 | { | |
3261 | return sock->ops->listen(sock, backlog); | |
3262 | } | |
c6d409cf | 3263 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e SS |
3264 | |
3265 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) | |
3266 | { | |
3267 | struct sock *sk = sock->sk; | |
3268 | int err; | |
3269 | ||
3270 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3271 | newsock); | |
3272 | if (err < 0) | |
3273 | goto done; | |
3274 | ||
cdfbabfb | 3275 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3276 | if (err < 0) { |
3277 | sock_release(*newsock); | |
fa8705b0 | 3278 | *newsock = NULL; |
ac5a488e SS |
3279 | goto done; |
3280 | } | |
3281 | ||
3282 | (*newsock)->ops = sock->ops; | |
1b08534e | 3283 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3284 | |
3285 | done: | |
3286 | return err; | |
3287 | } | |
c6d409cf | 3288 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e SS |
3289 | |
3290 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, | |
4768fbcb | 3291 | int flags) |
ac5a488e SS |
3292 | { |
3293 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3294 | } | |
c6d409cf | 3295 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e | 3296 | |
9b2c45d4 | 3297 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3298 | { |
9b2c45d4 | 3299 | return sock->ops->getname(sock, addr, 0); |
ac5a488e | 3300 | } |
c6d409cf | 3301 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e | 3302 | |
9b2c45d4 | 3303 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3304 | { |
9b2c45d4 | 3305 | return sock->ops->getname(sock, addr, 1); |
ac5a488e | 3306 | } |
c6d409cf | 3307 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e SS |
3308 | |
3309 | int kernel_getsockopt(struct socket *sock, int level, int optname, | |
3310 | char *optval, int *optlen) | |
3311 | { | |
3312 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3313 | char __user *uoptval; |
3314 | int __user *uoptlen; | |
ac5a488e SS |
3315 | int err; |
3316 | ||
fb8621bb NK |
3317 | uoptval = (char __user __force *) optval; |
3318 | uoptlen = (int __user __force *) optlen; | |
3319 | ||
ac5a488e SS |
3320 | set_fs(KERNEL_DS); |
3321 | if (level == SOL_SOCKET) | |
fb8621bb | 3322 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3323 | else |
fb8621bb NK |
3324 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3325 | uoptlen); | |
ac5a488e SS |
3326 | set_fs(oldfs); |
3327 | return err; | |
3328 | } | |
c6d409cf | 3329 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e SS |
3330 | |
3331 | int kernel_setsockopt(struct socket *sock, int level, int optname, | |
b7058842 | 3332 | char *optval, unsigned int optlen) |
ac5a488e SS |
3333 | { |
3334 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3335 | char __user *uoptval; |
ac5a488e SS |
3336 | int err; |
3337 | ||
fb8621bb NK |
3338 | uoptval = (char __user __force *) optval; |
3339 | ||
ac5a488e SS |
3340 | set_fs(KERNEL_DS); |
3341 | if (level == SOL_SOCKET) | |
fb8621bb | 3342 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3343 | else |
fb8621bb | 3344 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3345 | optlen); |
3346 | set_fs(oldfs); | |
3347 | return err; | |
3348 | } | |
c6d409cf | 3349 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e SS |
3350 | |
3351 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, | |
3352 | size_t size, int flags) | |
3353 | { | |
3354 | if (sock->ops->sendpage) | |
3355 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3356 | ||
3357 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3358 | } | |
c6d409cf | 3359 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3360 | |
306b13eb TH |
3361 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3362 | size_t size, int flags) | |
3363 | { | |
3364 | struct socket *sock = sk->sk_socket; | |
3365 | ||
3366 | if (sock->ops->sendpage_locked) | |
3367 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3368 | flags); | |
3369 | ||
3370 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3371 | } | |
3372 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3373 | ||
91cf45f0 TM |
3374 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3375 | { | |
3376 | return sock->ops->shutdown(sock, how); | |
3377 | } | |
91cf45f0 | 3378 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 P |
3379 | |
3380 | /* This routine returns the IP overhead imposed by a socket i.e. | |
3381 | * the length of the underlying IP header, depending on whether | |
3382 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
57240d00 | 3383 | * on at the socket. Assumes that the caller has a lock on the socket. |
113c3075 P |
3384 | */ |
3385 | u32 kernel_sock_ip_overhead(struct sock *sk) | |
3386 | { | |
3387 | struct inet_sock *inet; | |
3388 | struct ip_options_rcu *opt; | |
3389 | u32 overhead = 0; | |
113c3075 P |
3390 | #if IS_ENABLED(CONFIG_IPV6) |
3391 | struct ipv6_pinfo *np; | |
3392 | struct ipv6_txoptions *optv6 = NULL; | |
3393 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3394 | ||
3395 | if (!sk) | |
3396 | return overhead; | |
3397 | ||
113c3075 P |
3398 | switch (sk->sk_family) { |
3399 | case AF_INET: | |
3400 | inet = inet_sk(sk); | |
3401 | overhead += sizeof(struct iphdr); | |
3402 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3403 | sock_owned_by_user(sk)); |
113c3075 P |
3404 | if (opt) |
3405 | overhead += opt->opt.optlen; | |
3406 | return overhead; | |
3407 | #if IS_ENABLED(CONFIG_IPV6) | |
3408 | case AF_INET6: | |
3409 | np = inet6_sk(sk); | |
3410 | overhead += sizeof(struct ipv6hdr); | |
3411 | if (np) | |
3412 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3413 | sock_owned_by_user(sk)); |
113c3075 P |
3414 | if (optv6) |
3415 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3416 | return overhead; | |
3417 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3418 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3419 | return overhead; | |
3420 | } | |
3421 | } | |
3422 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |