]>
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> |
c8e8cd57 | 92 | #include <linux/nospec.h> |
1da177e4 | 93 | |
7c0f6ba6 | 94 | #include <linux/uaccess.h> |
1da177e4 LT |
95 | #include <asm/unistd.h> |
96 | ||
97 | #include <net/compat.h> | |
87de87d5 | 98 | #include <net/wext.h> |
f8451725 | 99 | #include <net/cls_cgroup.h> |
1da177e4 LT |
100 | |
101 | #include <net/sock.h> | |
102 | #include <linux/netfilter.h> | |
103 | ||
6b96018b AB |
104 | #include <linux/if_tun.h> |
105 | #include <linux/ipv6_route.h> | |
106 | #include <linux/route.h> | |
6b96018b | 107 | #include <linux/sockios.h> |
076bb0c8 | 108 | #include <net/busy_poll.h> |
f24b9be5 | 109 | #include <linux/errqueue.h> |
06021292 | 110 | |
e0d1095a | 111 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
112 | unsigned int sysctl_net_busy_read __read_mostly; |
113 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 114 | #endif |
6b96018b | 115 | |
8ae5e030 AV |
116 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
117 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 118 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
119 | |
120 | static int sock_close(struct inode *inode, struct file *file); | |
a11e1d43 LT |
121 | static __poll_t sock_poll(struct file *file, |
122 | struct poll_table_struct *wait); | |
89bddce5 | 123 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
124 | #ifdef CONFIG_COMPAT |
125 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 126 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 127 | #endif |
1da177e4 | 128 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
129 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
130 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 131 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 132 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 133 | unsigned int flags); |
1da177e4 | 134 | |
1da177e4 LT |
135 | /* |
136 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
137 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
138 | */ | |
139 | ||
da7071d7 | 140 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
141 | .owner = THIS_MODULE, |
142 | .llseek = no_llseek, | |
8ae5e030 AV |
143 | .read_iter = sock_read_iter, |
144 | .write_iter = sock_write_iter, | |
1da177e4 LT |
145 | .poll = sock_poll, |
146 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
147 | #ifdef CONFIG_COMPAT |
148 | .compat_ioctl = compat_sock_ioctl, | |
149 | #endif | |
1da177e4 | 150 | .mmap = sock_mmap, |
1da177e4 LT |
151 | .release = sock_close, |
152 | .fasync = sock_fasync, | |
5274f052 JA |
153 | .sendpage = sock_sendpage, |
154 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 155 | .splice_read = sock_splice_read, |
1da177e4 LT |
156 | }; |
157 | ||
158 | /* | |
159 | * The protocol list. Each protocol is registered in here. | |
160 | */ | |
161 | ||
1da177e4 | 162 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 163 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 164 | |
1da177e4 | 165 | /* |
89bddce5 SH |
166 | * Support routines. |
167 | * Move socket addresses back and forth across the kernel/user | |
168 | * divide and look after the messy bits. | |
1da177e4 LT |
169 | */ |
170 | ||
1da177e4 LT |
171 | /** |
172 | * move_addr_to_kernel - copy a socket address into kernel space | |
173 | * @uaddr: Address in user space | |
174 | * @kaddr: Address in kernel space | |
175 | * @ulen: Length in user space | |
176 | * | |
177 | * The address is copied into kernel space. If the provided address is | |
178 | * too long an error code of -EINVAL is returned. If the copy gives | |
179 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
180 | */ | |
181 | ||
43db362d | 182 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 183 | { |
230b1839 | 184 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 185 | return -EINVAL; |
89bddce5 | 186 | if (ulen == 0) |
1da177e4 | 187 | return 0; |
89bddce5 | 188 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 189 | return -EFAULT; |
3ec3b2fb | 190 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
191 | } |
192 | ||
193 | /** | |
194 | * move_addr_to_user - copy an address to user space | |
195 | * @kaddr: kernel space address | |
196 | * @klen: length of address in kernel | |
197 | * @uaddr: user space address | |
198 | * @ulen: pointer to user length field | |
199 | * | |
200 | * The value pointed to by ulen on entry is the buffer length available. | |
201 | * This is overwritten with the buffer space used. -EINVAL is returned | |
202 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
203 | * is returned if either the buffer or the length field are not | |
204 | * accessible. | |
205 | * After copying the data up to the limit the user specifies, the true | |
206 | * length of the data is written over the length limit the user | |
207 | * specified. Zero is returned for a success. | |
208 | */ | |
89bddce5 | 209 | |
43db362d | 210 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 211 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
212 | { |
213 | int err; | |
214 | int len; | |
215 | ||
68c6beb3 | 216 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
217 | err = get_user(len, ulen); |
218 | if (err) | |
1da177e4 | 219 | return err; |
89bddce5 SH |
220 | if (len > klen) |
221 | len = klen; | |
68c6beb3 | 222 | if (len < 0) |
1da177e4 | 223 | return -EINVAL; |
89bddce5 | 224 | if (len) { |
d6fe3945 SG |
225 | if (audit_sockaddr(klen, kaddr)) |
226 | return -ENOMEM; | |
89bddce5 | 227 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
228 | return -EFAULT; |
229 | } | |
230 | /* | |
89bddce5 SH |
231 | * "fromlen shall refer to the value before truncation.." |
232 | * 1003.1g | |
1da177e4 LT |
233 | */ |
234 | return __put_user(klen, ulen); | |
235 | } | |
236 | ||
08009a76 | 237 | static struct kmem_cache *sock_inode_cachep __ro_after_init; |
1da177e4 LT |
238 | |
239 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
240 | { | |
241 | struct socket_alloc *ei; | |
eaefd110 | 242 | struct socket_wq *wq; |
89bddce5 | 243 | |
e94b1766 | 244 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
245 | if (!ei) |
246 | return NULL; | |
eaefd110 ED |
247 | wq = kmalloc(sizeof(*wq), GFP_KERNEL); |
248 | if (!wq) { | |
43815482 ED |
249 | kmem_cache_free(sock_inode_cachep, ei); |
250 | return NULL; | |
251 | } | |
eaefd110 ED |
252 | init_waitqueue_head(&wq->wait); |
253 | wq->fasync_list = NULL; | |
574aab1e | 254 | wq->flags = 0; |
e6476c21 | 255 | ei->socket.wq = wq; |
89bddce5 | 256 | |
1da177e4 LT |
257 | ei->socket.state = SS_UNCONNECTED; |
258 | ei->socket.flags = 0; | |
259 | ei->socket.ops = NULL; | |
260 | ei->socket.sk = NULL; | |
261 | ei->socket.file = NULL; | |
1da177e4 LT |
262 | |
263 | return &ei->vfs_inode; | |
264 | } | |
265 | ||
266 | static void sock_destroy_inode(struct inode *inode) | |
267 | { | |
43815482 ED |
268 | struct socket_alloc *ei; |
269 | ||
270 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
e6476c21 | 271 | kfree_rcu(ei->socket.wq, rcu); |
43815482 | 272 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
273 | } |
274 | ||
51cc5068 | 275 | static void init_once(void *foo) |
1da177e4 | 276 | { |
89bddce5 | 277 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 278 | |
a35afb83 | 279 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 280 | } |
89bddce5 | 281 | |
1e911632 | 282 | static void init_inodecache(void) |
1da177e4 LT |
283 | { |
284 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
285 | sizeof(struct socket_alloc), |
286 | 0, | |
287 | (SLAB_HWCACHE_ALIGN | | |
288 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 289 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 290 | init_once); |
1e911632 | 291 | BUG_ON(sock_inode_cachep == NULL); |
1da177e4 LT |
292 | } |
293 | ||
b87221de | 294 | static const struct super_operations sockfs_ops = { |
c6d409cf ED |
295 | .alloc_inode = sock_alloc_inode, |
296 | .destroy_inode = sock_destroy_inode, | |
297 | .statfs = simple_statfs, | |
1da177e4 LT |
298 | }; |
299 | ||
c23fbb6b ED |
300 | /* |
301 | * sockfs_dname() is called from d_path(). | |
302 | */ | |
303 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
304 | { | |
305 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 306 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
307 | } |
308 | ||
3ba13d17 | 309 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 310 | .d_dname = sockfs_dname, |
1da177e4 LT |
311 | }; |
312 | ||
bba0bd31 AG |
313 | static int sockfs_xattr_get(const struct xattr_handler *handler, |
314 | struct dentry *dentry, struct inode *inode, | |
315 | const char *suffix, void *value, size_t size) | |
316 | { | |
317 | if (value) { | |
318 | if (dentry->d_name.len + 1 > size) | |
319 | return -ERANGE; | |
320 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
321 | } | |
322 | return dentry->d_name.len + 1; | |
323 | } | |
324 | ||
325 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" | |
326 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
327 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
328 | ||
329 | static const struct xattr_handler sockfs_xattr_handler = { | |
330 | .name = XATTR_NAME_SOCKPROTONAME, | |
331 | .get = sockfs_xattr_get, | |
332 | }; | |
333 | ||
4a590153 AG |
334 | static int sockfs_security_xattr_set(const struct xattr_handler *handler, |
335 | struct dentry *dentry, struct inode *inode, | |
336 | const char *suffix, const void *value, | |
337 | size_t size, int flags) | |
338 | { | |
339 | /* Handled by LSM. */ | |
340 | return -EAGAIN; | |
341 | } | |
342 | ||
343 | static const struct xattr_handler sockfs_security_xattr_handler = { | |
344 | .prefix = XATTR_SECURITY_PREFIX, | |
345 | .set = sockfs_security_xattr_set, | |
346 | }; | |
347 | ||
bba0bd31 AG |
348 | static const struct xattr_handler *sockfs_xattr_handlers[] = { |
349 | &sockfs_xattr_handler, | |
4a590153 | 350 | &sockfs_security_xattr_handler, |
bba0bd31 AG |
351 | NULL |
352 | }; | |
353 | ||
c74a1cbb AV |
354 | static struct dentry *sockfs_mount(struct file_system_type *fs_type, |
355 | int flags, const char *dev_name, void *data) | |
356 | { | |
bba0bd31 AG |
357 | return mount_pseudo_xattr(fs_type, "socket:", &sockfs_ops, |
358 | sockfs_xattr_handlers, | |
359 | &sockfs_dentry_operations, SOCKFS_MAGIC); | |
c74a1cbb AV |
360 | } |
361 | ||
362 | static struct vfsmount *sock_mnt __read_mostly; | |
363 | ||
364 | static struct file_system_type sock_fs_type = { | |
365 | .name = "sockfs", | |
366 | .mount = sockfs_mount, | |
367 | .kill_sb = kill_anon_super, | |
368 | }; | |
369 | ||
1da177e4 LT |
370 | /* |
371 | * Obtains the first available file descriptor and sets it up for use. | |
372 | * | |
39d8c1b6 DM |
373 | * These functions create file structures and maps them to fd space |
374 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
375 | * and file struct implicitly stored in sock->file. |
376 | * Note that another thread may close file descriptor before we return | |
377 | * from this function. We use the fact that now we do not refer | |
378 | * to socket after mapping. If one day we will need it, this | |
379 | * function will increment ref. count on file by 1. | |
380 | * | |
381 | * In any case returned fd MAY BE not valid! | |
382 | * This race condition is unavoidable | |
383 | * with shared fd spaces, we cannot solve it inside kernel, | |
384 | * but we take care of internal coherence yet. | |
385 | */ | |
386 | ||
aab174f0 | 387 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 388 | { |
7cbe66b6 | 389 | struct qstr name = { .name = "" }; |
2c48b9c4 | 390 | struct path path; |
7cbe66b6 | 391 | struct file *file; |
1da177e4 | 392 | |
600e1779 MY |
393 | if (dname) { |
394 | name.name = dname; | |
395 | name.len = strlen(name.name); | |
396 | } else if (sock->sk) { | |
397 | name.name = sock->sk->sk_prot_creator->name; | |
398 | name.len = strlen(name.name); | |
399 | } | |
4b936885 | 400 | path.dentry = d_alloc_pseudo(sock_mnt->mnt_sb, &name); |
8e1611e2 AV |
401 | if (unlikely(!path.dentry)) { |
402 | sock_release(sock); | |
28407630 | 403 | return ERR_PTR(-ENOMEM); |
8e1611e2 | 404 | } |
2c48b9c4 | 405 | path.mnt = mntget(sock_mnt); |
39d8c1b6 | 406 | |
2c48b9c4 | 407 | d_instantiate(path.dentry, SOCK_INODE(sock)); |
39d8c1b6 | 408 | |
2c48b9c4 | 409 | file = alloc_file(&path, FMODE_READ | FMODE_WRITE, |
ce8d2cdf | 410 | &socket_file_ops); |
b5ffe634 | 411 | if (IS_ERR(file)) { |
8e1611e2 | 412 | /* drop dentry, keep inode for a bit */ |
c5ef6035 | 413 | ihold(d_inode(path.dentry)); |
2c48b9c4 | 414 | path_put(&path); |
8e1611e2 AV |
415 | /* ... and now kill it properly */ |
416 | sock_release(sock); | |
39b65252 | 417 | return file; |
cc3808f8 AV |
418 | } |
419 | ||
420 | sock->file = file; | |
77d27200 | 421 | file->f_flags = O_RDWR | (flags & O_NONBLOCK); |
39d8c1b6 | 422 | file->private_data = sock; |
28407630 | 423 | return file; |
39d8c1b6 | 424 | } |
56b31d1c | 425 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 426 | |
56b31d1c | 427 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
428 | { |
429 | struct file *newfile; | |
28407630 | 430 | int fd = get_unused_fd_flags(flags); |
ce4bb04c AV |
431 | if (unlikely(fd < 0)) { |
432 | sock_release(sock); | |
28407630 | 433 | return fd; |
ce4bb04c | 434 | } |
39d8c1b6 | 435 | |
aab174f0 | 436 | newfile = sock_alloc_file(sock, flags, NULL); |
28407630 | 437 | if (likely(!IS_ERR(newfile))) { |
39d8c1b6 | 438 | fd_install(fd, newfile); |
28407630 AV |
439 | return fd; |
440 | } | |
7cbe66b6 | 441 | |
28407630 AV |
442 | put_unused_fd(fd); |
443 | return PTR_ERR(newfile); | |
1da177e4 LT |
444 | } |
445 | ||
406a3c63 | 446 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 447 | { |
6cb153ca BL |
448 | if (file->f_op == &socket_file_ops) |
449 | return file->private_data; /* set in sock_map_fd */ | |
450 | ||
23bb80d2 ED |
451 | *err = -ENOTSOCK; |
452 | return NULL; | |
6cb153ca | 453 | } |
406a3c63 | 454 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 455 | |
1da177e4 | 456 | /** |
c6d409cf | 457 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
458 | * @fd: file handle |
459 | * @err: pointer to an error code return | |
460 | * | |
461 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 462 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
463 | * with a negative errno code and NULL is returned. The function checks |
464 | * for both invalid handles and passing a handle which is not a socket. | |
465 | * | |
466 | * On a success the socket object pointer is returned. | |
467 | */ | |
468 | ||
469 | struct socket *sockfd_lookup(int fd, int *err) | |
470 | { | |
471 | struct file *file; | |
1da177e4 LT |
472 | struct socket *sock; |
473 | ||
89bddce5 SH |
474 | file = fget(fd); |
475 | if (!file) { | |
1da177e4 LT |
476 | *err = -EBADF; |
477 | return NULL; | |
478 | } | |
89bddce5 | 479 | |
6cb153ca BL |
480 | sock = sock_from_file(file, err); |
481 | if (!sock) | |
1da177e4 | 482 | fput(file); |
6cb153ca BL |
483 | return sock; |
484 | } | |
c6d409cf | 485 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 486 | |
6cb153ca BL |
487 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
488 | { | |
00e188ef | 489 | struct fd f = fdget(fd); |
6cb153ca BL |
490 | struct socket *sock; |
491 | ||
3672558c | 492 | *err = -EBADF; |
00e188ef AV |
493 | if (f.file) { |
494 | sock = sock_from_file(f.file, err); | |
495 | if (likely(sock)) { | |
496 | *fput_needed = f.flags; | |
6cb153ca | 497 | return sock; |
00e188ef AV |
498 | } |
499 | fdput(f); | |
1da177e4 | 500 | } |
6cb153ca | 501 | return NULL; |
1da177e4 LT |
502 | } |
503 | ||
600e1779 MY |
504 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
505 | size_t size) | |
506 | { | |
507 | ssize_t len; | |
508 | ssize_t used = 0; | |
509 | ||
c5ef6035 | 510 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
511 | if (len < 0) |
512 | return len; | |
513 | used += len; | |
514 | if (buffer) { | |
515 | if (size < used) | |
516 | return -ERANGE; | |
517 | buffer += len; | |
518 | } | |
519 | ||
520 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
521 | used += len; | |
522 | if (buffer) { | |
523 | if (size < used) | |
524 | return -ERANGE; | |
525 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
526 | buffer += len; | |
527 | } | |
528 | ||
529 | return used; | |
530 | } | |
531 | ||
dc647ec8 | 532 | static int sockfs_setattr(struct dentry *dentry, struct iattr *iattr) |
86741ec2 LC |
533 | { |
534 | int err = simple_setattr(dentry, iattr); | |
535 | ||
e1a3a60a | 536 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
537 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
538 | ||
6d8c50dc CW |
539 | if (sock->sk) |
540 | sock->sk->sk_uid = iattr->ia_uid; | |
541 | else | |
542 | err = -ENOENT; | |
86741ec2 LC |
543 | } |
544 | ||
545 | return err; | |
546 | } | |
547 | ||
600e1779 | 548 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 549 | .listxattr = sockfs_listxattr, |
86741ec2 | 550 | .setattr = sockfs_setattr, |
600e1779 MY |
551 | }; |
552 | ||
1da177e4 LT |
553 | /** |
554 | * sock_alloc - allocate a socket | |
89bddce5 | 555 | * |
1da177e4 LT |
556 | * Allocate a new inode and socket object. The two are bound together |
557 | * and initialised. The socket is then returned. If we are out of inodes | |
558 | * NULL is returned. | |
559 | */ | |
560 | ||
f4a00aac | 561 | struct socket *sock_alloc(void) |
1da177e4 | 562 | { |
89bddce5 SH |
563 | struct inode *inode; |
564 | struct socket *sock; | |
1da177e4 | 565 | |
a209dfc7 | 566 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
567 | if (!inode) |
568 | return NULL; | |
569 | ||
570 | sock = SOCKET_I(inode); | |
571 | ||
85fe4025 | 572 | inode->i_ino = get_next_ino(); |
89bddce5 | 573 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
574 | inode->i_uid = current_fsuid(); |
575 | inode->i_gid = current_fsgid(); | |
600e1779 | 576 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 577 | |
1da177e4 LT |
578 | return sock; |
579 | } | |
f4a00aac | 580 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 581 | |
1da177e4 LT |
582 | /** |
583 | * sock_release - close a socket | |
584 | * @sock: socket to close | |
585 | * | |
586 | * The socket is released from the protocol stack if it has a release | |
587 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 588 | * an inode not a file. |
1da177e4 | 589 | */ |
89bddce5 | 590 | |
6d8c50dc | 591 | static void __sock_release(struct socket *sock, struct inode *inode) |
1da177e4 LT |
592 | { |
593 | if (sock->ops) { | |
594 | struct module *owner = sock->ops->owner; | |
595 | ||
6d8c50dc CW |
596 | if (inode) |
597 | inode_lock(inode); | |
1da177e4 | 598 | sock->ops->release(sock); |
6d8c50dc CW |
599 | if (inode) |
600 | inode_unlock(inode); | |
1da177e4 LT |
601 | sock->ops = NULL; |
602 | module_put(owner); | |
603 | } | |
604 | ||
e6476c21 | 605 | if (sock->wq->fasync_list) |
3410f22e | 606 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 607 | |
1da177e4 LT |
608 | if (!sock->file) { |
609 | iput(SOCK_INODE(sock)); | |
610 | return; | |
611 | } | |
89bddce5 | 612 | sock->file = NULL; |
1da177e4 | 613 | } |
6d8c50dc CW |
614 | |
615 | void sock_release(struct socket *sock) | |
616 | { | |
617 | __sock_release(sock, NULL); | |
618 | } | |
c6d409cf | 619 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 620 | |
c14ac945 | 621 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 622 | { |
140c55d4 ED |
623 | u8 flags = *tx_flags; |
624 | ||
c14ac945 | 625 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
626 | flags |= SKBTX_HW_TSTAMP; |
627 | ||
c14ac945 | 628 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
629 | flags |= SKBTX_SW_TSTAMP; |
630 | ||
c14ac945 | 631 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
632 | flags |= SKBTX_SCHED_TSTAMP; |
633 | ||
140c55d4 | 634 | *tx_flags = flags; |
20d49473 | 635 | } |
67cc0d40 | 636 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 637 | |
d8725c86 | 638 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 639 | { |
01e97e65 | 640 | int ret = sock->ops->sendmsg(sock, msg, msg_data_left(msg)); |
d8725c86 AV |
641 | BUG_ON(ret == -EIOCBQUEUED); |
642 | return ret; | |
1da177e4 LT |
643 | } |
644 | ||
d8725c86 | 645 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 646 | { |
d8725c86 | 647 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 648 | msg_data_left(msg)); |
228e548e | 649 | |
d8725c86 | 650 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 651 | } |
c6d409cf | 652 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 LT |
653 | |
654 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, | |
655 | struct kvec *vec, size_t num, size_t size) | |
656 | { | |
6aa24814 | 657 | iov_iter_kvec(&msg->msg_iter, WRITE | ITER_KVEC, vec, num, size); |
d8725c86 | 658 | return sock_sendmsg(sock, msg); |
1da177e4 | 659 | } |
c6d409cf | 660 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 661 | |
306b13eb TH |
662 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
663 | struct kvec *vec, size_t num, size_t size) | |
664 | { | |
665 | struct socket *sock = sk->sk_socket; | |
666 | ||
667 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 668 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb TH |
669 | |
670 | iov_iter_kvec(&msg->msg_iter, WRITE | ITER_KVEC, vec, num, size); | |
671 | ||
672 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
673 | } | |
674 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
675 | ||
8605330a SHY |
676 | static bool skb_is_err_queue(const struct sk_buff *skb) |
677 | { | |
678 | /* pkt_type of skbs enqueued on the error queue are set to | |
679 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
680 | * in recvmsg, since skbs received on a local socket will never | |
681 | * have a pkt_type of PACKET_OUTGOING. | |
682 | */ | |
683 | return skb->pkt_type == PACKET_OUTGOING; | |
684 | } | |
685 | ||
b50a5c70 ML |
686 | /* On transmit, software and hardware timestamps are returned independently. |
687 | * As the two skb clones share the hardware timestamp, which may be updated | |
688 | * before the software timestamp is received, a hardware TX timestamp may be | |
689 | * returned only if there is no software TX timestamp. Ignore false software | |
690 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
691 | * option SO_TIMESTAMP(NS) is enabled on the socket, even when the skb has a | |
692 | * hardware timestamp. | |
693 | */ | |
694 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
695 | { | |
696 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
697 | } | |
698 | ||
aad9c8c4 ML |
699 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb) |
700 | { | |
701 | struct scm_ts_pktinfo ts_pktinfo; | |
702 | struct net_device *orig_dev; | |
703 | ||
704 | if (!skb_mac_header_was_set(skb)) | |
705 | return; | |
706 | ||
707 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
708 | ||
709 | rcu_read_lock(); | |
710 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
711 | if (orig_dev) | |
712 | ts_pktinfo.if_index = orig_dev->ifindex; | |
713 | rcu_read_unlock(); | |
714 | ||
715 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
716 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
717 | sizeof(ts_pktinfo), &ts_pktinfo); | |
718 | } | |
719 | ||
92f37fd2 ED |
720 | /* |
721 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
722 | */ | |
723 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
724 | struct sk_buff *skb) | |
725 | { | |
20d49473 | 726 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
f24b9be5 | 727 | struct scm_timestamping tss; |
b50a5c70 | 728 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
729 | struct skb_shared_hwtstamps *shhwtstamps = |
730 | skb_hwtstamps(skb); | |
731 | ||
732 | /* Race occurred between timestamp enabling and packet | |
733 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 734 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 735 | __net_timestamp(skb); |
b50a5c70 ML |
736 | false_tstamp = 1; |
737 | } | |
20d49473 PO |
738 | |
739 | if (need_software_tstamp) { | |
740 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
741 | struct timeval tv; | |
742 | skb_get_timestamp(skb, &tv); | |
743 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMP, | |
744 | sizeof(tv), &tv); | |
745 | } else { | |
f24b9be5 WB |
746 | struct timespec ts; |
747 | skb_get_timestampns(skb, &ts); | |
20d49473 | 748 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPNS, |
f24b9be5 | 749 | sizeof(ts), &ts); |
20d49473 PO |
750 | } |
751 | } | |
752 | ||
f24b9be5 | 753 | memset(&tss, 0, sizeof(tss)); |
c199105d | 754 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
f24b9be5 | 755 | ktime_to_timespec_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 756 | empty = 0; |
4d276eb6 | 757 | if (shhwtstamps && |
b9f40e21 | 758 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
b50a5c70 | 759 | !skb_is_swtx_tstamp(skb, false_tstamp) && |
aad9c8c4 | 760 | ktime_to_timespec_cond(shhwtstamps->hwtstamp, tss.ts + 2)) { |
4d276eb6 | 761 | empty = 0; |
aad9c8c4 ML |
762 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && |
763 | !skb_is_err_queue(skb)) | |
764 | put_ts_pktinfo(msg, skb); | |
765 | } | |
1c885808 | 766 | if (!empty) { |
20d49473 | 767 | put_cmsg(msg, SOL_SOCKET, |
f24b9be5 | 768 | SCM_TIMESTAMPING, sizeof(tss), &tss); |
1c885808 | 769 | |
8605330a | 770 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 771 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
772 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
773 | skb->len, skb->data); | |
774 | } | |
92f37fd2 | 775 | } |
7c81fd8b ACM |
776 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
777 | ||
6e3e939f JB |
778 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
779 | struct sk_buff *skb) | |
780 | { | |
781 | int ack; | |
782 | ||
783 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
784 | return; | |
785 | if (!skb->wifi_acked_valid) | |
786 | return; | |
787 | ||
788 | ack = skb->wifi_acked; | |
789 | ||
790 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
791 | } | |
792 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
793 | ||
11165f14 | 794 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
795 | struct sk_buff *skb) | |
3b885787 | 796 | { |
744d5a3e | 797 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 798 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 799 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
800 | } |
801 | ||
767dd033 | 802 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
803 | struct sk_buff *skb) |
804 | { | |
805 | sock_recv_timestamp(msg, sk, skb); | |
806 | sock_recv_drops(msg, sk, skb); | |
807 | } | |
767dd033 | 808 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 809 | |
1b784140 | 810 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 811 | int flags) |
1da177e4 | 812 | { |
2da62906 | 813 | return sock->ops->recvmsg(sock, msg, msg_data_left(msg), flags); |
1da177e4 LT |
814 | } |
815 | ||
2da62906 | 816 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 817 | { |
2da62906 | 818 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 819 | |
2da62906 | 820 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 821 | } |
c6d409cf | 822 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 823 | |
c1249c0a ML |
824 | /** |
825 | * kernel_recvmsg - Receive a message from a socket (kernel space) | |
826 | * @sock: The socket to receive the message from | |
827 | * @msg: Received message | |
828 | * @vec: Input s/g array for message data | |
829 | * @num: Size of input s/g array | |
830 | * @size: Number of bytes to read | |
831 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
832 | * | |
833 | * On return the msg structure contains the scatter/gather array passed in the | |
834 | * vec argument. The array is modified so that it consists of the unfilled | |
835 | * portion of the original array. | |
836 | * | |
837 | * The returned value is the total number of bytes received, or an error. | |
838 | */ | |
89bddce5 SH |
839 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
840 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
841 | { |
842 | mm_segment_t oldfs = get_fs(); | |
843 | int result; | |
844 | ||
6aa24814 | 845 | iov_iter_kvec(&msg->msg_iter, READ | ITER_KVEC, vec, num, size); |
1da177e4 | 846 | set_fs(KERNEL_DS); |
2da62906 | 847 | result = sock_recvmsg(sock, msg, flags); |
1da177e4 LT |
848 | set_fs(oldfs); |
849 | return result; | |
850 | } | |
c6d409cf | 851 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 852 | |
ce1d4d3e CH |
853 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
854 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 855 | { |
1da177e4 LT |
856 | struct socket *sock; |
857 | int flags; | |
858 | ||
ce1d4d3e CH |
859 | sock = file->private_data; |
860 | ||
35f9c09f ED |
861 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
862 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
863 | flags |= more; | |
ce1d4d3e | 864 | |
e6949583 | 865 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 866 | } |
1da177e4 | 867 | |
9c55e01c | 868 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 869 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
870 | unsigned int flags) |
871 | { | |
872 | struct socket *sock = file->private_data; | |
873 | ||
997b37da RDC |
874 | if (unlikely(!sock->ops->splice_read)) |
875 | return -EINVAL; | |
876 | ||
9c55e01c JA |
877 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
878 | } | |
879 | ||
8ae5e030 | 880 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 881 | { |
6d652330 AV |
882 | struct file *file = iocb->ki_filp; |
883 | struct socket *sock = file->private_data; | |
0345f931 | 884 | struct msghdr msg = {.msg_iter = *to, |
885 | .msg_iocb = iocb}; | |
8ae5e030 | 886 | ssize_t res; |
ce1d4d3e | 887 | |
8ae5e030 AV |
888 | if (file->f_flags & O_NONBLOCK) |
889 | msg.msg_flags = MSG_DONTWAIT; | |
890 | ||
891 | if (iocb->ki_pos != 0) | |
1da177e4 | 892 | return -ESPIPE; |
027445c3 | 893 | |
66ee59af | 894 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
895 | return 0; |
896 | ||
2da62906 | 897 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
898 | *to = msg.msg_iter; |
899 | return res; | |
1da177e4 LT |
900 | } |
901 | ||
8ae5e030 | 902 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 903 | { |
6d652330 AV |
904 | struct file *file = iocb->ki_filp; |
905 | struct socket *sock = file->private_data; | |
0345f931 | 906 | struct msghdr msg = {.msg_iter = *from, |
907 | .msg_iocb = iocb}; | |
8ae5e030 | 908 | ssize_t res; |
1da177e4 | 909 | |
8ae5e030 | 910 | if (iocb->ki_pos != 0) |
ce1d4d3e | 911 | return -ESPIPE; |
027445c3 | 912 | |
8ae5e030 AV |
913 | if (file->f_flags & O_NONBLOCK) |
914 | msg.msg_flags = MSG_DONTWAIT; | |
915 | ||
6d652330 AV |
916 | if (sock->type == SOCK_SEQPACKET) |
917 | msg.msg_flags |= MSG_EOR; | |
918 | ||
d8725c86 | 919 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
920 | *from = msg.msg_iter; |
921 | return res; | |
1da177e4 LT |
922 | } |
923 | ||
1da177e4 LT |
924 | /* |
925 | * Atomic setting of ioctl hooks to avoid race | |
926 | * with module unload. | |
927 | */ | |
928 | ||
4a3e2f71 | 929 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 930 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 931 | |
881d966b | 932 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 933 | { |
4a3e2f71 | 934 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 935 | br_ioctl_hook = hook; |
4a3e2f71 | 936 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
937 | } |
938 | EXPORT_SYMBOL(brioctl_set); | |
939 | ||
4a3e2f71 | 940 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 941 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 942 | |
881d966b | 943 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 944 | { |
4a3e2f71 | 945 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 946 | vlan_ioctl_hook = hook; |
4a3e2f71 | 947 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
948 | } |
949 | EXPORT_SYMBOL(vlan_ioctl_set); | |
950 | ||
4a3e2f71 | 951 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 952 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 953 | |
89bddce5 | 954 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 955 | { |
4a3e2f71 | 956 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 957 | dlci_ioctl_hook = hook; |
4a3e2f71 | 958 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
959 | } |
960 | EXPORT_SYMBOL(dlci_ioctl_set); | |
961 | ||
6b96018b AB |
962 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
963 | unsigned int cmd, unsigned long arg) | |
964 | { | |
965 | int err; | |
966 | void __user *argp = (void __user *)arg; | |
967 | ||
968 | err = sock->ops->ioctl(sock, cmd, arg); | |
969 | ||
970 | /* | |
971 | * If this ioctl is unknown try to hand it down | |
972 | * to the NIC driver. | |
973 | */ | |
36fd633e AV |
974 | if (err != -ENOIOCTLCMD) |
975 | return err; | |
6b96018b | 976 | |
36fd633e AV |
977 | if (cmd == SIOCGIFCONF) { |
978 | struct ifconf ifc; | |
979 | if (copy_from_user(&ifc, argp, sizeof(struct ifconf))) | |
980 | return -EFAULT; | |
981 | rtnl_lock(); | |
982 | err = dev_ifconf(net, &ifc, sizeof(struct ifreq)); | |
983 | rtnl_unlock(); | |
984 | if (!err && copy_to_user(argp, &ifc, sizeof(struct ifconf))) | |
985 | err = -EFAULT; | |
44c02a2c AV |
986 | } else { |
987 | struct ifreq ifr; | |
988 | bool need_copyout; | |
989 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
990 | return -EFAULT; | |
991 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
992 | if (!err && need_copyout) | |
993 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
994 | return -EFAULT; | |
36fd633e | 995 | } |
6b96018b AB |
996 | return err; |
997 | } | |
998 | ||
1da177e4 LT |
999 | /* |
1000 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1001 | * what to do with it - that's up to the protocol still. | |
1002 | */ | |
1003 | ||
d8d211a2 | 1004 | struct ns_common *get_net_ns(struct ns_common *ns) |
c62cce2c AV |
1005 | { |
1006 | return &get_net(container_of(ns, struct net, ns))->ns; | |
1007 | } | |
d8d211a2 | 1008 | EXPORT_SYMBOL_GPL(get_net_ns); |
c62cce2c | 1009 | |
1da177e4 LT |
1010 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
1011 | { | |
1012 | struct socket *sock; | |
881d966b | 1013 | struct sock *sk; |
1da177e4 LT |
1014 | void __user *argp = (void __user *)arg; |
1015 | int pid, err; | |
881d966b | 1016 | struct net *net; |
1da177e4 | 1017 | |
b69aee04 | 1018 | sock = file->private_data; |
881d966b | 1019 | sk = sock->sk; |
3b1e0a65 | 1020 | net = sock_net(sk); |
44c02a2c AV |
1021 | if (unlikely(cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))) { |
1022 | struct ifreq ifr; | |
1023 | bool need_copyout; | |
1024 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
1025 | return -EFAULT; | |
1026 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1027 | if (!err && need_copyout) | |
1028 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
1029 | return -EFAULT; | |
1da177e4 | 1030 | } else |
3d23e349 | 1031 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1032 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
b1b0c245 | 1033 | err = wext_handle_ioctl(net, cmd, argp); |
1da177e4 | 1034 | } else |
3d23e349 | 1035 | #endif |
89bddce5 | 1036 | switch (cmd) { |
1da177e4 LT |
1037 | case FIOSETOWN: |
1038 | case SIOCSPGRP: | |
1039 | err = -EFAULT; | |
1040 | if (get_user(pid, (int __user *)argp)) | |
1041 | break; | |
393cc3f5 | 1042 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1043 | break; |
1044 | case FIOGETOWN: | |
1045 | case SIOCGPGRP: | |
609d7fa9 | 1046 | err = put_user(f_getown(sock->file), |
89bddce5 | 1047 | (int __user *)argp); |
1da177e4 LT |
1048 | break; |
1049 | case SIOCGIFBR: | |
1050 | case SIOCSIFBR: | |
1051 | case SIOCBRADDBR: | |
1052 | case SIOCBRDELBR: | |
1053 | err = -ENOPKG; | |
1054 | if (!br_ioctl_hook) | |
1055 | request_module("bridge"); | |
1056 | ||
4a3e2f71 | 1057 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1058 | if (br_ioctl_hook) |
881d966b | 1059 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1060 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1061 | break; |
1062 | case SIOCGIFVLAN: | |
1063 | case SIOCSIFVLAN: | |
1064 | err = -ENOPKG; | |
1065 | if (!vlan_ioctl_hook) | |
1066 | request_module("8021q"); | |
1067 | ||
4a3e2f71 | 1068 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1069 | if (vlan_ioctl_hook) |
881d966b | 1070 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1071 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1072 | break; |
1da177e4 LT |
1073 | case SIOCADDDLCI: |
1074 | case SIOCDELDLCI: | |
1075 | err = -ENOPKG; | |
1076 | if (!dlci_ioctl_hook) | |
1077 | request_module("dlci"); | |
1078 | ||
7512cbf6 PE |
1079 | mutex_lock(&dlci_ioctl_mutex); |
1080 | if (dlci_ioctl_hook) | |
1da177e4 | 1081 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1082 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 | 1083 | break; |
c62cce2c AV |
1084 | case SIOCGSKNS: |
1085 | err = -EPERM; | |
1086 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1087 | break; | |
1088 | ||
1089 | err = open_related_ns(&net->ns, get_net_ns); | |
1090 | break; | |
1da177e4 | 1091 | default: |
6b96018b | 1092 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1093 | break; |
89bddce5 | 1094 | } |
1da177e4 LT |
1095 | return err; |
1096 | } | |
1097 | ||
1098 | int sock_create_lite(int family, int type, int protocol, struct socket **res) | |
1099 | { | |
1100 | int err; | |
1101 | struct socket *sock = NULL; | |
89bddce5 | 1102 | |
1da177e4 LT |
1103 | err = security_socket_create(family, type, protocol, 1); |
1104 | if (err) | |
1105 | goto out; | |
1106 | ||
1107 | sock = sock_alloc(); | |
1108 | if (!sock) { | |
1109 | err = -ENOMEM; | |
1110 | goto out; | |
1111 | } | |
1112 | ||
1da177e4 | 1113 | sock->type = type; |
7420ed23 VY |
1114 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1115 | if (err) | |
1116 | goto out_release; | |
1117 | ||
1da177e4 LT |
1118 | out: |
1119 | *res = sock; | |
1120 | return err; | |
7420ed23 VY |
1121 | out_release: |
1122 | sock_release(sock); | |
1123 | sock = NULL; | |
1124 | goto out; | |
1da177e4 | 1125 | } |
c6d409cf | 1126 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1127 | |
1128 | /* No kernel lock held - perfect */ | |
ade994f4 | 1129 | static __poll_t sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1130 | { |
3cafb376 | 1131 | struct socket *sock = file->private_data; |
a331de3b | 1132 | __poll_t events = poll_requested_events(wait), flag = 0; |
2d48d67f | 1133 | |
e88958e6 CH |
1134 | if (!sock->ops->poll) |
1135 | return 0; | |
f641f13b | 1136 | |
a331de3b CH |
1137 | if (sk_can_busy_loop(sock->sk)) { |
1138 | /* poll once if requested by the syscall */ | |
1139 | if (events & POLL_BUSY_LOOP) | |
1140 | sk_busy_loop(sock->sk, 1); | |
1141 | ||
1142 | /* if this socket can poll_ll, tell the system call */ | |
1143 | flag = POLL_BUSY_LOOP; | |
1144 | } | |
1145 | ||
1146 | return sock->ops->poll(file, sock, wait) | flag; | |
1da177e4 LT |
1147 | } |
1148 | ||
89bddce5 | 1149 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1150 | { |
b69aee04 | 1151 | struct socket *sock = file->private_data; |
1da177e4 LT |
1152 | |
1153 | return sock->ops->mmap(file, sock, vma); | |
1154 | } | |
1155 | ||
20380731 | 1156 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1157 | { |
6d8c50dc | 1158 | __sock_release(SOCKET_I(inode), inode); |
1da177e4 LT |
1159 | return 0; |
1160 | } | |
1161 | ||
1162 | /* | |
1163 | * Update the socket async list | |
1164 | * | |
1165 | * Fasync_list locking strategy. | |
1166 | * | |
1167 | * 1. fasync_list is modified only under process context socket lock | |
1168 | * i.e. under semaphore. | |
1169 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1170 | * or under socket lock |
1da177e4 LT |
1171 | */ |
1172 | ||
1173 | static int sock_fasync(int fd, struct file *filp, int on) | |
1174 | { | |
989a2979 ED |
1175 | struct socket *sock = filp->private_data; |
1176 | struct sock *sk = sock->sk; | |
eaefd110 | 1177 | struct socket_wq *wq; |
1da177e4 | 1178 | |
989a2979 | 1179 | if (sk == NULL) |
1da177e4 | 1180 | return -EINVAL; |
1da177e4 LT |
1181 | |
1182 | lock_sock(sk); | |
e6476c21 | 1183 | wq = sock->wq; |
eaefd110 | 1184 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1185 | |
eaefd110 | 1186 | if (!wq->fasync_list) |
989a2979 ED |
1187 | sock_reset_flag(sk, SOCK_FASYNC); |
1188 | else | |
bcdce719 | 1189 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1190 | |
989a2979 | 1191 | release_sock(sk); |
1da177e4 LT |
1192 | return 0; |
1193 | } | |
1194 | ||
ceb5d58b | 1195 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1196 | |
ceb5d58b | 1197 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1198 | { |
ceb5d58b | 1199 | if (!wq || !wq->fasync_list) |
1da177e4 | 1200 | return -1; |
ceb5d58b | 1201 | |
89bddce5 | 1202 | switch (how) { |
8d8ad9d7 | 1203 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1204 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1205 | break; |
1206 | goto call_kill; | |
8d8ad9d7 | 1207 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1208 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 LT |
1209 | break; |
1210 | /* fall through */ | |
8d8ad9d7 | 1211 | case SOCK_WAKE_IO: |
89bddce5 | 1212 | call_kill: |
43815482 | 1213 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1214 | break; |
8d8ad9d7 | 1215 | case SOCK_WAKE_URG: |
43815482 | 1216 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1217 | } |
ceb5d58b | 1218 | |
1da177e4 LT |
1219 | return 0; |
1220 | } | |
c6d409cf | 1221 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1222 | |
721db93a | 1223 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1224 | struct socket **res, int kern) |
1da177e4 LT |
1225 | { |
1226 | int err; | |
1227 | struct socket *sock; | |
55737fda | 1228 | const struct net_proto_family *pf; |
1da177e4 LT |
1229 | |
1230 | /* | |
89bddce5 | 1231 | * Check protocol is in range |
1da177e4 LT |
1232 | */ |
1233 | if (family < 0 || family >= NPROTO) | |
1234 | return -EAFNOSUPPORT; | |
1235 | if (type < 0 || type >= SOCK_MAX) | |
1236 | return -EINVAL; | |
1237 | ||
1238 | /* Compatibility. | |
1239 | ||
1240 | This uglymoron is moved from INET layer to here to avoid | |
1241 | deadlock in module load. | |
1242 | */ | |
1243 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1244 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1245 | current->comm); | |
1da177e4 LT |
1246 | family = PF_PACKET; |
1247 | } | |
1248 | ||
1249 | err = security_socket_create(family, type, protocol, kern); | |
1250 | if (err) | |
1251 | return err; | |
89bddce5 | 1252 | |
55737fda SH |
1253 | /* |
1254 | * Allocate the socket and allow the family to set things up. if | |
1255 | * the protocol is 0, the family is instructed to select an appropriate | |
1256 | * default. | |
1257 | */ | |
1258 | sock = sock_alloc(); | |
1259 | if (!sock) { | |
e87cc472 | 1260 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1261 | return -ENFILE; /* Not exactly a match, but its the |
1262 | closest posix thing */ | |
1263 | } | |
1264 | ||
1265 | sock->type = type; | |
1266 | ||
95a5afca | 1267 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1268 | /* Attempt to load a protocol module if the find failed. |
1269 | * | |
1270 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1271 | * requested real, full-featured networking support upon configuration. |
1272 | * Otherwise module support will break! | |
1273 | */ | |
190683a9 | 1274 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1275 | request_module("net-pf-%d", family); |
1da177e4 LT |
1276 | #endif |
1277 | ||
55737fda SH |
1278 | rcu_read_lock(); |
1279 | pf = rcu_dereference(net_families[family]); | |
1280 | err = -EAFNOSUPPORT; | |
1281 | if (!pf) | |
1282 | goto out_release; | |
1da177e4 LT |
1283 | |
1284 | /* | |
1285 | * We will call the ->create function, that possibly is in a loadable | |
1286 | * module, so we have to bump that loadable module refcnt first. | |
1287 | */ | |
55737fda | 1288 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1289 | goto out_release; |
1290 | ||
55737fda SH |
1291 | /* Now protected by module ref count */ |
1292 | rcu_read_unlock(); | |
1293 | ||
3f378b68 | 1294 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1295 | if (err < 0) |
1da177e4 | 1296 | goto out_module_put; |
a79af59e | 1297 | |
1da177e4 LT |
1298 | /* |
1299 | * Now to bump the refcnt of the [loadable] module that owns this | |
1300 | * socket at sock_release time we decrement its refcnt. | |
1301 | */ | |
55737fda SH |
1302 | if (!try_module_get(sock->ops->owner)) |
1303 | goto out_module_busy; | |
1304 | ||
1da177e4 LT |
1305 | /* |
1306 | * Now that we're done with the ->create function, the [loadable] | |
1307 | * module can have its refcnt decremented | |
1308 | */ | |
55737fda | 1309 | module_put(pf->owner); |
7420ed23 VY |
1310 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1311 | if (err) | |
3b185525 | 1312 | goto out_sock_release; |
55737fda | 1313 | *res = sock; |
1da177e4 | 1314 | |
55737fda SH |
1315 | return 0; |
1316 | ||
1317 | out_module_busy: | |
1318 | err = -EAFNOSUPPORT; | |
1da177e4 | 1319 | out_module_put: |
55737fda SH |
1320 | sock->ops = NULL; |
1321 | module_put(pf->owner); | |
1322 | out_sock_release: | |
1da177e4 | 1323 | sock_release(sock); |
55737fda SH |
1324 | return err; |
1325 | ||
1326 | out_release: | |
1327 | rcu_read_unlock(); | |
1328 | goto out_sock_release; | |
1da177e4 | 1329 | } |
721db93a | 1330 | EXPORT_SYMBOL(__sock_create); |
1da177e4 LT |
1331 | |
1332 | int sock_create(int family, int type, int protocol, struct socket **res) | |
1333 | { | |
1b8d7ae4 | 1334 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1335 | } |
c6d409cf | 1336 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1337 | |
eeb1bd5c | 1338 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1339 | { |
eeb1bd5c | 1340 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1341 | } |
c6d409cf | 1342 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1343 | |
9d6a15c3 | 1344 | int __sys_socket(int family, int type, int protocol) |
1da177e4 LT |
1345 | { |
1346 | int retval; | |
1347 | struct socket *sock; | |
a677a039 UD |
1348 | int flags; |
1349 | ||
e38b36f3 UD |
1350 | /* Check the SOCK_* constants for consistency. */ |
1351 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1352 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1353 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1354 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1355 | ||
a677a039 | 1356 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1357 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1358 | return -EINVAL; |
1359 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1360 | |
aaca0bdc UD |
1361 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1362 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1363 | ||
1da177e4 LT |
1364 | retval = sock_create(family, type, protocol, &sock); |
1365 | if (retval < 0) | |
8e1611e2 | 1366 | return retval; |
1da177e4 | 1367 | |
8e1611e2 | 1368 | return sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1369 | } |
1370 | ||
9d6a15c3 DB |
1371 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1372 | { | |
1373 | return __sys_socket(family, type, protocol); | |
1374 | } | |
1375 | ||
1da177e4 LT |
1376 | /* |
1377 | * Create a pair of connected sockets. | |
1378 | */ | |
1379 | ||
6debc8d8 | 1380 | int __sys_socketpair(int family, int type, int protocol, int __user *usockvec) |
1da177e4 LT |
1381 | { |
1382 | struct socket *sock1, *sock2; | |
1383 | int fd1, fd2, err; | |
db349509 | 1384 | struct file *newfile1, *newfile2; |
a677a039 UD |
1385 | int flags; |
1386 | ||
1387 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1388 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1389 | return -EINVAL; |
1390 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1391 | |
aaca0bdc UD |
1392 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1393 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1394 | ||
016a266b AV |
1395 | /* |
1396 | * reserve descriptors and make sure we won't fail | |
1397 | * to return them to userland. | |
1398 | */ | |
1399 | fd1 = get_unused_fd_flags(flags); | |
1400 | if (unlikely(fd1 < 0)) | |
1401 | return fd1; | |
1402 | ||
1403 | fd2 = get_unused_fd_flags(flags); | |
1404 | if (unlikely(fd2 < 0)) { | |
1405 | put_unused_fd(fd1); | |
1406 | return fd2; | |
1407 | } | |
1408 | ||
1409 | err = put_user(fd1, &usockvec[0]); | |
1410 | if (err) | |
1411 | goto out; | |
1412 | ||
1413 | err = put_user(fd2, &usockvec[1]); | |
1414 | if (err) | |
1415 | goto out; | |
1416 | ||
1da177e4 LT |
1417 | /* |
1418 | * Obtain the first socket and check if the underlying protocol | |
1419 | * supports the socketpair call. | |
1420 | */ | |
1421 | ||
1422 | err = sock_create(family, type, protocol, &sock1); | |
016a266b | 1423 | if (unlikely(err < 0)) |
1da177e4 LT |
1424 | goto out; |
1425 | ||
1426 | err = sock_create(family, type, protocol, &sock2); | |
016a266b AV |
1427 | if (unlikely(err < 0)) { |
1428 | sock_release(sock1); | |
1429 | goto out; | |
bf3c23d1 | 1430 | } |
d73aa286 | 1431 | |
d47cd945 DH |
1432 | err = security_socket_socketpair(sock1, sock2); |
1433 | if (unlikely(err)) { | |
1434 | sock_release(sock2); | |
1435 | sock_release(sock1); | |
1436 | goto out; | |
1437 | } | |
1438 | ||
016a266b AV |
1439 | err = sock1->ops->socketpair(sock1, sock2); |
1440 | if (unlikely(err < 0)) { | |
1441 | sock_release(sock2); | |
1442 | sock_release(sock1); | |
1443 | goto out; | |
28407630 AV |
1444 | } |
1445 | ||
aab174f0 | 1446 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1447 | if (IS_ERR(newfile1)) { |
28407630 | 1448 | err = PTR_ERR(newfile1); |
016a266b AV |
1449 | sock_release(sock2); |
1450 | goto out; | |
28407630 AV |
1451 | } |
1452 | ||
aab174f0 | 1453 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1454 | if (IS_ERR(newfile2)) { |
1455 | err = PTR_ERR(newfile2); | |
016a266b AV |
1456 | fput(newfile1); |
1457 | goto out; | |
db349509 AV |
1458 | } |
1459 | ||
157cf649 | 1460 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1461 | |
db349509 AV |
1462 | fd_install(fd1, newfile1); |
1463 | fd_install(fd2, newfile2); | |
d73aa286 | 1464 | return 0; |
1da177e4 | 1465 | |
016a266b | 1466 | out: |
d73aa286 | 1467 | put_unused_fd(fd2); |
d73aa286 | 1468 | put_unused_fd(fd1); |
1da177e4 LT |
1469 | return err; |
1470 | } | |
1471 | ||
6debc8d8 DB |
1472 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1473 | int __user *, usockvec) | |
1474 | { | |
1475 | return __sys_socketpair(family, type, protocol, usockvec); | |
1476 | } | |
1477 | ||
1da177e4 LT |
1478 | /* |
1479 | * Bind a name to a socket. Nothing much to do here since it's | |
1480 | * the protocol's responsibility to handle the local address. | |
1481 | * | |
1482 | * We move the socket address to kernel space before we call | |
1483 | * the protocol layer (having also checked the address is ok). | |
1484 | */ | |
1485 | ||
a87d35d8 | 1486 | int __sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
1da177e4 LT |
1487 | { |
1488 | struct socket *sock; | |
230b1839 | 1489 | struct sockaddr_storage address; |
6cb153ca | 1490 | int err, fput_needed; |
1da177e4 | 1491 | |
89bddce5 | 1492 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1493 | if (sock) { |
43db362d | 1494 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
89bddce5 SH |
1495 | if (err >= 0) { |
1496 | err = security_socket_bind(sock, | |
230b1839 | 1497 | (struct sockaddr *)&address, |
89bddce5 | 1498 | addrlen); |
6cb153ca BL |
1499 | if (!err) |
1500 | err = sock->ops->bind(sock, | |
89bddce5 | 1501 | (struct sockaddr *) |
230b1839 | 1502 | &address, addrlen); |
1da177e4 | 1503 | } |
6cb153ca | 1504 | fput_light(sock->file, fput_needed); |
89bddce5 | 1505 | } |
1da177e4 LT |
1506 | return err; |
1507 | } | |
1508 | ||
a87d35d8 DB |
1509 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1510 | { | |
1511 | return __sys_bind(fd, umyaddr, addrlen); | |
1512 | } | |
1513 | ||
1da177e4 LT |
1514 | /* |
1515 | * Perform a listen. Basically, we allow the protocol to do anything | |
1516 | * necessary for a listen, and if that works, we mark the socket as | |
1517 | * ready for listening. | |
1518 | */ | |
1519 | ||
25e290ee | 1520 | int __sys_listen(int fd, int backlog) |
1da177e4 LT |
1521 | { |
1522 | struct socket *sock; | |
6cb153ca | 1523 | int err, fput_needed; |
b8e1f9b5 | 1524 | int somaxconn; |
89bddce5 SH |
1525 | |
1526 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1527 | if (sock) { | |
8efa6e93 | 1528 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1529 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1530 | backlog = somaxconn; |
1da177e4 LT |
1531 | |
1532 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1533 | if (!err) |
1534 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1535 | |
6cb153ca | 1536 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1537 | } |
1538 | return err; | |
1539 | } | |
1540 | ||
25e290ee DB |
1541 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1542 | { | |
1543 | return __sys_listen(fd, backlog); | |
1544 | } | |
1545 | ||
1da177e4 LT |
1546 | /* |
1547 | * For accept, we attempt to create a new socket, set up the link | |
1548 | * with the client, wake up the client, then return the new | |
1549 | * connected fd. We collect the address of the connector in kernel | |
1550 | * space and move it to user at the very end. This is unclean because | |
1551 | * we open the socket then return an error. | |
1552 | * | |
1553 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1554 | * status to recvmsg. We need to add that support in a way thats | |
b903036a | 1555 | * clean when we restructure accept also. |
1da177e4 LT |
1556 | */ |
1557 | ||
4541e805 DB |
1558 | int __sys_accept4(int fd, struct sockaddr __user *upeer_sockaddr, |
1559 | int __user *upeer_addrlen, int flags) | |
1da177e4 LT |
1560 | { |
1561 | struct socket *sock, *newsock; | |
39d8c1b6 | 1562 | struct file *newfile; |
6cb153ca | 1563 | int err, len, newfd, fput_needed; |
230b1839 | 1564 | struct sockaddr_storage address; |
1da177e4 | 1565 | |
77d27200 | 1566 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1567 | return -EINVAL; |
1568 | ||
1569 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1570 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1571 | ||
6cb153ca | 1572 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1573 | if (!sock) |
1574 | goto out; | |
1575 | ||
1576 | err = -ENFILE; | |
c6d409cf ED |
1577 | newsock = sock_alloc(); |
1578 | if (!newsock) | |
1da177e4 LT |
1579 | goto out_put; |
1580 | ||
1581 | newsock->type = sock->type; | |
1582 | newsock->ops = sock->ops; | |
1583 | ||
1da177e4 LT |
1584 | /* |
1585 | * We don't need try_module_get here, as the listening socket (sock) | |
1586 | * has the protocol module (sock->ops->owner) held. | |
1587 | */ | |
1588 | __module_get(newsock->ops->owner); | |
1589 | ||
28407630 | 1590 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1591 | if (unlikely(newfd < 0)) { |
1592 | err = newfd; | |
9a1875e6 DM |
1593 | sock_release(newsock); |
1594 | goto out_put; | |
39d8c1b6 | 1595 | } |
aab174f0 | 1596 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1597 | if (IS_ERR(newfile)) { |
28407630 AV |
1598 | err = PTR_ERR(newfile); |
1599 | put_unused_fd(newfd); | |
28407630 AV |
1600 | goto out_put; |
1601 | } | |
39d8c1b6 | 1602 | |
a79af59e FF |
1603 | err = security_socket_accept(sock, newsock); |
1604 | if (err) | |
39d8c1b6 | 1605 | goto out_fd; |
a79af59e | 1606 | |
cdfbabfb | 1607 | err = sock->ops->accept(sock, newsock, sock->file->f_flags, false); |
1da177e4 | 1608 | if (err < 0) |
39d8c1b6 | 1609 | goto out_fd; |
1da177e4 LT |
1610 | |
1611 | if (upeer_sockaddr) { | |
9b2c45d4 DV |
1612 | len = newsock->ops->getname(newsock, |
1613 | (struct sockaddr *)&address, 2); | |
1614 | if (len < 0) { | |
1da177e4 | 1615 | err = -ECONNABORTED; |
39d8c1b6 | 1616 | goto out_fd; |
1da177e4 | 1617 | } |
43db362d | 1618 | err = move_addr_to_user(&address, |
230b1839 | 1619 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1620 | if (err < 0) |
39d8c1b6 | 1621 | goto out_fd; |
1da177e4 LT |
1622 | } |
1623 | ||
1624 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1625 | ||
39d8c1b6 DM |
1626 | fd_install(newfd, newfile); |
1627 | err = newfd; | |
1da177e4 | 1628 | |
1da177e4 | 1629 | out_put: |
6cb153ca | 1630 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1631 | out: |
1632 | return err; | |
39d8c1b6 | 1633 | out_fd: |
9606a216 | 1634 | fput(newfile); |
39d8c1b6 | 1635 | put_unused_fd(newfd); |
1da177e4 LT |
1636 | goto out_put; |
1637 | } | |
1638 | ||
4541e805 DB |
1639 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1640 | int __user *, upeer_addrlen, int, flags) | |
1641 | { | |
1642 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, flags); | |
1643 | } | |
1644 | ||
20f37034 HC |
1645 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1646 | int __user *, upeer_addrlen) | |
aaca0bdc | 1647 | { |
4541e805 | 1648 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1649 | } |
1650 | ||
1da177e4 LT |
1651 | /* |
1652 | * Attempt to connect to a socket with the server address. The address | |
1653 | * is in user space so we verify it is OK and move it to kernel space. | |
1654 | * | |
1655 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1656 | * break bindings | |
1657 | * | |
1658 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1659 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1660 | * include the -EINPROGRESS status for such sockets. | |
1661 | */ | |
1662 | ||
1387c2c2 | 1663 | int __sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
1da177e4 LT |
1664 | { |
1665 | struct socket *sock; | |
230b1839 | 1666 | struct sockaddr_storage address; |
6cb153ca | 1667 | int err, fput_needed; |
1da177e4 | 1668 | |
6cb153ca | 1669 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1670 | if (!sock) |
1671 | goto out; | |
43db362d | 1672 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1673 | if (err < 0) |
1674 | goto out_put; | |
1675 | ||
89bddce5 | 1676 | err = |
230b1839 | 1677 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1678 | if (err) |
1679 | goto out_put; | |
1680 | ||
230b1839 | 1681 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1682 | sock->file->f_flags); |
1683 | out_put: | |
6cb153ca | 1684 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1685 | out: |
1686 | return err; | |
1687 | } | |
1688 | ||
1387c2c2 DB |
1689 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1690 | int, addrlen) | |
1691 | { | |
1692 | return __sys_connect(fd, uservaddr, addrlen); | |
1693 | } | |
1694 | ||
1da177e4 LT |
1695 | /* |
1696 | * Get the local address ('name') of a socket object. Move the obtained | |
1697 | * name to user space. | |
1698 | */ | |
1699 | ||
8882a107 DB |
1700 | int __sys_getsockname(int fd, struct sockaddr __user *usockaddr, |
1701 | int __user *usockaddr_len) | |
1da177e4 LT |
1702 | { |
1703 | struct socket *sock; | |
230b1839 | 1704 | struct sockaddr_storage address; |
9b2c45d4 | 1705 | int err, fput_needed; |
89bddce5 | 1706 | |
6cb153ca | 1707 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1708 | if (!sock) |
1709 | goto out; | |
1710 | ||
1711 | err = security_socket_getsockname(sock); | |
1712 | if (err) | |
1713 | goto out_put; | |
1714 | ||
9b2c45d4 DV |
1715 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 0); |
1716 | if (err < 0) | |
1da177e4 | 1717 | goto out_put; |
9b2c45d4 DV |
1718 | /* "err" is actually length in this case */ |
1719 | err = move_addr_to_user(&address, err, usockaddr, usockaddr_len); | |
1da177e4 LT |
1720 | |
1721 | out_put: | |
6cb153ca | 1722 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1723 | out: |
1724 | return err; | |
1725 | } | |
1726 | ||
8882a107 DB |
1727 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1728 | int __user *, usockaddr_len) | |
1729 | { | |
1730 | return __sys_getsockname(fd, usockaddr, usockaddr_len); | |
1731 | } | |
1732 | ||
1da177e4 LT |
1733 | /* |
1734 | * Get the remote address ('name') of a socket object. Move the obtained | |
1735 | * name to user space. | |
1736 | */ | |
1737 | ||
b21c8f83 DB |
1738 | int __sys_getpeername(int fd, struct sockaddr __user *usockaddr, |
1739 | int __user *usockaddr_len) | |
1da177e4 LT |
1740 | { |
1741 | struct socket *sock; | |
230b1839 | 1742 | struct sockaddr_storage address; |
9b2c45d4 | 1743 | int err, fput_needed; |
1da177e4 | 1744 | |
89bddce5 SH |
1745 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1746 | if (sock != NULL) { | |
1da177e4 LT |
1747 | err = security_socket_getpeername(sock); |
1748 | if (err) { | |
6cb153ca | 1749 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1750 | return err; |
1751 | } | |
1752 | ||
9b2c45d4 DV |
1753 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 1); |
1754 | if (err >= 0) | |
1755 | /* "err" is actually length in this case */ | |
1756 | err = move_addr_to_user(&address, err, usockaddr, | |
89bddce5 | 1757 | usockaddr_len); |
6cb153ca | 1758 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1759 | } |
1760 | return err; | |
1761 | } | |
1762 | ||
b21c8f83 DB |
1763 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1764 | int __user *, usockaddr_len) | |
1765 | { | |
1766 | return __sys_getpeername(fd, usockaddr, usockaddr_len); | |
1767 | } | |
1768 | ||
1da177e4 LT |
1769 | /* |
1770 | * Send a datagram to a given address. We move the address into kernel | |
1771 | * space and check the user space data area is readable before invoking | |
1772 | * the protocol. | |
1773 | */ | |
211b634b DB |
1774 | int __sys_sendto(int fd, void __user *buff, size_t len, unsigned int flags, |
1775 | struct sockaddr __user *addr, int addr_len) | |
1da177e4 LT |
1776 | { |
1777 | struct socket *sock; | |
230b1839 | 1778 | struct sockaddr_storage address; |
1da177e4 LT |
1779 | int err; |
1780 | struct msghdr msg; | |
1781 | struct iovec iov; | |
6cb153ca | 1782 | int fput_needed; |
6cb153ca | 1783 | |
602bd0e9 AV |
1784 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1785 | if (unlikely(err)) | |
1786 | return err; | |
de0fa95c PE |
1787 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1788 | if (!sock) | |
4387ff75 | 1789 | goto out; |
6cb153ca | 1790 | |
89bddce5 | 1791 | msg.msg_name = NULL; |
89bddce5 SH |
1792 | msg.msg_control = NULL; |
1793 | msg.msg_controllen = 0; | |
1794 | msg.msg_namelen = 0; | |
6cb153ca | 1795 | if (addr) { |
43db362d | 1796 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1797 | if (err < 0) |
1798 | goto out_put; | |
230b1839 | 1799 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1800 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1801 | } |
1802 | if (sock->file->f_flags & O_NONBLOCK) | |
1803 | flags |= MSG_DONTWAIT; | |
1804 | msg.msg_flags = flags; | |
d8725c86 | 1805 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 1806 | |
89bddce5 | 1807 | out_put: |
de0fa95c | 1808 | fput_light(sock->file, fput_needed); |
4387ff75 | 1809 | out: |
1da177e4 LT |
1810 | return err; |
1811 | } | |
1812 | ||
211b634b DB |
1813 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
1814 | unsigned int, flags, struct sockaddr __user *, addr, | |
1815 | int, addr_len) | |
1816 | { | |
1817 | return __sys_sendto(fd, buff, len, flags, addr, addr_len); | |
1818 | } | |
1819 | ||
1da177e4 | 1820 | /* |
89bddce5 | 1821 | * Send a datagram down a socket. |
1da177e4 LT |
1822 | */ |
1823 | ||
3e0fa65f | 1824 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1825 | unsigned int, flags) |
1da177e4 | 1826 | { |
211b634b | 1827 | return __sys_sendto(fd, buff, len, flags, NULL, 0); |
1da177e4 LT |
1828 | } |
1829 | ||
1830 | /* | |
89bddce5 | 1831 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1832 | * sender. We verify the buffers are writable and if needed move the |
1833 | * sender address from kernel to user space. | |
1834 | */ | |
7a09e1eb DB |
1835 | int __sys_recvfrom(int fd, void __user *ubuf, size_t size, unsigned int flags, |
1836 | struct sockaddr __user *addr, int __user *addr_len) | |
1da177e4 LT |
1837 | { |
1838 | struct socket *sock; | |
1839 | struct iovec iov; | |
1840 | struct msghdr msg; | |
230b1839 | 1841 | struct sockaddr_storage address; |
89bddce5 | 1842 | int err, err2; |
6cb153ca BL |
1843 | int fput_needed; |
1844 | ||
602bd0e9 AV |
1845 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
1846 | if (unlikely(err)) | |
1847 | return err; | |
de0fa95c | 1848 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 1849 | if (!sock) |
de0fa95c | 1850 | goto out; |
1da177e4 | 1851 | |
89bddce5 SH |
1852 | msg.msg_control = NULL; |
1853 | msg.msg_controllen = 0; | |
f3d33426 HFS |
1854 | /* Save some cycles and don't copy the address if not needed */ |
1855 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
1856 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
1857 | msg.msg_namelen = 0; | |
130ed5d1 | 1858 | msg.msg_iocb = NULL; |
9f138fa6 | 1859 | msg.msg_flags = 0; |
1da177e4 LT |
1860 | if (sock->file->f_flags & O_NONBLOCK) |
1861 | flags |= MSG_DONTWAIT; | |
2da62906 | 1862 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 1863 | |
89bddce5 | 1864 | if (err >= 0 && addr != NULL) { |
43db362d | 1865 | err2 = move_addr_to_user(&address, |
230b1839 | 1866 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
1867 | if (err2 < 0) |
1868 | err = err2; | |
1da177e4 | 1869 | } |
de0fa95c PE |
1870 | |
1871 | fput_light(sock->file, fput_needed); | |
4387ff75 | 1872 | out: |
1da177e4 LT |
1873 | return err; |
1874 | } | |
1875 | ||
7a09e1eb DB |
1876 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
1877 | unsigned int, flags, struct sockaddr __user *, addr, | |
1878 | int __user *, addr_len) | |
1879 | { | |
1880 | return __sys_recvfrom(fd, ubuf, size, flags, addr, addr_len); | |
1881 | } | |
1882 | ||
1da177e4 | 1883 | /* |
89bddce5 | 1884 | * Receive a datagram from a socket. |
1da177e4 LT |
1885 | */ |
1886 | ||
b7c0ddf5 JG |
1887 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
1888 | unsigned int, flags) | |
1da177e4 | 1889 | { |
7a09e1eb | 1890 | return __sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
1da177e4 LT |
1891 | } |
1892 | ||
1893 | /* | |
1894 | * Set a socket option. Because we don't know the option lengths we have | |
1895 | * to pass the user mode parameter for the protocols to sort out. | |
1896 | */ | |
1897 | ||
cc36dca0 DB |
1898 | static int __sys_setsockopt(int fd, int level, int optname, |
1899 | char __user *optval, int optlen) | |
1da177e4 | 1900 | { |
6cb153ca | 1901 | int err, fput_needed; |
1da177e4 LT |
1902 | struct socket *sock; |
1903 | ||
1904 | if (optlen < 0) | |
1905 | return -EINVAL; | |
89bddce5 SH |
1906 | |
1907 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1908 | if (sock != NULL) { | |
1909 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
1910 | if (err) |
1911 | goto out_put; | |
1da177e4 LT |
1912 | |
1913 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1914 | err = |
1915 | sock_setsockopt(sock, level, optname, optval, | |
1916 | optlen); | |
1da177e4 | 1917 | else |
89bddce5 SH |
1918 | err = |
1919 | sock->ops->setsockopt(sock, level, optname, optval, | |
1920 | optlen); | |
6cb153ca BL |
1921 | out_put: |
1922 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1923 | } |
1924 | return err; | |
1925 | } | |
1926 | ||
cc36dca0 DB |
1927 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
1928 | char __user *, optval, int, optlen) | |
1929 | { | |
1930 | return __sys_setsockopt(fd, level, optname, optval, optlen); | |
1931 | } | |
1932 | ||
1da177e4 LT |
1933 | /* |
1934 | * Get a socket option. Because we don't know the option lengths we have | |
1935 | * to pass a user mode parameter for the protocols to sort out. | |
1936 | */ | |
1937 | ||
13a2d70e DB |
1938 | static int __sys_getsockopt(int fd, int level, int optname, |
1939 | char __user *optval, int __user *optlen) | |
1da177e4 | 1940 | { |
6cb153ca | 1941 | int err, fput_needed; |
1da177e4 LT |
1942 | struct socket *sock; |
1943 | ||
89bddce5 SH |
1944 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1945 | if (sock != NULL) { | |
6cb153ca BL |
1946 | err = security_socket_getsockopt(sock, level, optname); |
1947 | if (err) | |
1948 | goto out_put; | |
1da177e4 LT |
1949 | |
1950 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1951 | err = |
1952 | sock_getsockopt(sock, level, optname, optval, | |
1953 | optlen); | |
1da177e4 | 1954 | else |
89bddce5 SH |
1955 | err = |
1956 | sock->ops->getsockopt(sock, level, optname, optval, | |
1957 | optlen); | |
6cb153ca BL |
1958 | out_put: |
1959 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1960 | } |
1961 | return err; | |
1962 | } | |
1963 | ||
13a2d70e DB |
1964 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
1965 | char __user *, optval, int __user *, optlen) | |
1966 | { | |
1967 | return __sys_getsockopt(fd, level, optname, optval, optlen); | |
1968 | } | |
1969 | ||
1da177e4 LT |
1970 | /* |
1971 | * Shutdown a socket. | |
1972 | */ | |
1973 | ||
005a1aea | 1974 | int __sys_shutdown(int fd, int how) |
1da177e4 | 1975 | { |
6cb153ca | 1976 | int err, fput_needed; |
1da177e4 LT |
1977 | struct socket *sock; |
1978 | ||
89bddce5 SH |
1979 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1980 | if (sock != NULL) { | |
1da177e4 | 1981 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
1982 | if (!err) |
1983 | err = sock->ops->shutdown(sock, how); | |
1984 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1985 | } |
1986 | return err; | |
1987 | } | |
1988 | ||
005a1aea DB |
1989 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
1990 | { | |
1991 | return __sys_shutdown(fd, how); | |
1992 | } | |
1993 | ||
89bddce5 | 1994 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
1995 | * fields which are the same type (int / unsigned) on our platforms. |
1996 | */ | |
1997 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
1998 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
1999 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
2000 | ||
c71d8ebe TH |
2001 | struct used_address { |
2002 | struct sockaddr_storage name; | |
2003 | unsigned int name_len; | |
2004 | }; | |
2005 | ||
da184284 AV |
2006 | static int copy_msghdr_from_user(struct msghdr *kmsg, |
2007 | struct user_msghdr __user *umsg, | |
2008 | struct sockaddr __user **save_addr, | |
2009 | struct iovec **iov) | |
1661bf36 | 2010 | { |
ffb07550 | 2011 | struct user_msghdr msg; |
08adb7da AV |
2012 | ssize_t err; |
2013 | ||
ffb07550 | 2014 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 2015 | return -EFAULT; |
dbb490b9 | 2016 | |
864d9664 | 2017 | kmsg->msg_control = (void __force *)msg.msg_control; |
ffb07550 AV |
2018 | kmsg->msg_controllen = msg.msg_controllen; |
2019 | kmsg->msg_flags = msg.msg_flags; | |
2020 | ||
2021 | kmsg->msg_namelen = msg.msg_namelen; | |
2022 | if (!msg.msg_name) | |
6a2a2b3a AS |
2023 | kmsg->msg_namelen = 0; |
2024 | ||
dbb490b9 ML |
2025 | if (kmsg->msg_namelen < 0) |
2026 | return -EINVAL; | |
2027 | ||
1661bf36 | 2028 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 2029 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
2030 | |
2031 | if (save_addr) | |
ffb07550 | 2032 | *save_addr = msg.msg_name; |
08adb7da | 2033 | |
ffb07550 | 2034 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 2035 | if (!save_addr) { |
864d9664 PA |
2036 | err = move_addr_to_kernel(msg.msg_name, |
2037 | kmsg->msg_namelen, | |
08adb7da AV |
2038 | kmsg->msg_name); |
2039 | if (err < 0) | |
2040 | return err; | |
2041 | } | |
2042 | } else { | |
2043 | kmsg->msg_name = NULL; | |
2044 | kmsg->msg_namelen = 0; | |
2045 | } | |
2046 | ||
ffb07550 | 2047 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
2048 | return -EMSGSIZE; |
2049 | ||
0345f931 | 2050 | kmsg->msg_iocb = NULL; |
2051 | ||
ffb07550 AV |
2052 | return import_iovec(save_addr ? READ : WRITE, |
2053 | msg.msg_iov, msg.msg_iovlen, | |
da184284 | 2054 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
1661bf36 DC |
2055 | } |
2056 | ||
666547ff | 2057 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2058 | struct msghdr *msg_sys, unsigned int flags, |
28a94d8f TH |
2059 | struct used_address *used_address, |
2060 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 2061 | { |
89bddce5 SH |
2062 | struct compat_msghdr __user *msg_compat = |
2063 | (struct compat_msghdr __user *)msg; | |
230b1839 | 2064 | struct sockaddr_storage address; |
1da177e4 | 2065 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 2066 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 2067 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 2068 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 2069 | unsigned char *ctl_buf = ctl; |
d8725c86 | 2070 | int ctl_len; |
08adb7da | 2071 | ssize_t err; |
89bddce5 | 2072 | |
08adb7da | 2073 | msg_sys->msg_name = &address; |
1da177e4 | 2074 | |
08449320 | 2075 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2076 | err = get_compat_msghdr(msg_sys, msg_compat, NULL, &iov); |
08449320 | 2077 | else |
08adb7da | 2078 | err = copy_msghdr_from_user(msg_sys, msg, NULL, &iov); |
89bddce5 | 2079 | if (err < 0) |
da184284 | 2080 | return err; |
1da177e4 LT |
2081 | |
2082 | err = -ENOBUFS; | |
2083 | ||
228e548e | 2084 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 2085 | goto out_freeiov; |
28a94d8f | 2086 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2087 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2088 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2089 | err = |
228e548e | 2090 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2091 | sizeof(ctl)); |
1da177e4 LT |
2092 | if (err) |
2093 | goto out_freeiov; | |
228e548e AB |
2094 | ctl_buf = msg_sys->msg_control; |
2095 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2096 | } else if (ctl_len) { |
ac4340fc DM |
2097 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2098 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2099 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2100 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2101 | if (ctl_buf == NULL) |
1da177e4 LT |
2102 | goto out_freeiov; |
2103 | } | |
2104 | err = -EFAULT; | |
2105 | /* | |
228e548e | 2106 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2107 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2108 | * checking falls down on this. | |
2109 | */ | |
fb8621bb | 2110 | if (copy_from_user(ctl_buf, |
228e548e | 2111 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2112 | ctl_len)) |
1da177e4 | 2113 | goto out_freectl; |
228e548e | 2114 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2115 | } |
228e548e | 2116 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2117 | |
2118 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2119 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2120 | /* |
2121 | * If this is sendmmsg() and current destination address is same as | |
2122 | * previously succeeded address, omit asking LSM's decision. | |
2123 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2124 | * destination address never matches. | |
2125 | */ | |
bc909d9d MD |
2126 | if (used_address && msg_sys->msg_name && |
2127 | used_address->name_len == msg_sys->msg_namelen && | |
2128 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2129 | used_address->name_len)) { |
d8725c86 | 2130 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2131 | goto out_freectl; |
2132 | } | |
d8725c86 | 2133 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2134 | /* |
2135 | * If this is sendmmsg() and sending to current destination address was | |
2136 | * successful, remember it. | |
2137 | */ | |
2138 | if (used_address && err >= 0) { | |
2139 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2140 | if (msg_sys->msg_name) |
2141 | memcpy(&used_address->name, msg_sys->msg_name, | |
2142 | used_address->name_len); | |
c71d8ebe | 2143 | } |
1da177e4 LT |
2144 | |
2145 | out_freectl: | |
89bddce5 | 2146 | if (ctl_buf != ctl) |
1da177e4 LT |
2147 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
2148 | out_freeiov: | |
da184284 | 2149 | kfree(iov); |
228e548e AB |
2150 | return err; |
2151 | } | |
2152 | ||
2153 | /* | |
2154 | * BSD sendmsg interface | |
2155 | */ | |
2156 | ||
e1834a32 DB |
2157 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2158 | bool forbid_cmsg_compat) | |
228e548e AB |
2159 | { |
2160 | int fput_needed, err; | |
2161 | struct msghdr msg_sys; | |
1be374a0 AL |
2162 | struct socket *sock; |
2163 | ||
e1834a32 DB |
2164 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2165 | return -EINVAL; | |
2166 | ||
1be374a0 | 2167 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2168 | if (!sock) |
2169 | goto out; | |
2170 | ||
28a94d8f | 2171 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2172 | |
6cb153ca | 2173 | fput_light(sock->file, fput_needed); |
89bddce5 | 2174 | out: |
1da177e4 LT |
2175 | return err; |
2176 | } | |
2177 | ||
666547ff | 2178 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 | 2179 | { |
e1834a32 | 2180 | return __sys_sendmsg(fd, msg, flags, true); |
a7526eb5 AL |
2181 | } |
2182 | ||
228e548e AB |
2183 | /* |
2184 | * Linux sendmmsg interface | |
2185 | */ | |
2186 | ||
2187 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
e1834a32 | 2188 | unsigned int flags, bool forbid_cmsg_compat) |
228e548e AB |
2189 | { |
2190 | int fput_needed, err, datagrams; | |
2191 | struct socket *sock; | |
2192 | struct mmsghdr __user *entry; | |
2193 | struct compat_mmsghdr __user *compat_entry; | |
2194 | struct msghdr msg_sys; | |
c71d8ebe | 2195 | struct used_address used_address; |
f092276d | 2196 | unsigned int oflags = flags; |
228e548e | 2197 | |
e1834a32 DB |
2198 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2199 | return -EINVAL; | |
2200 | ||
98382f41 AB |
2201 | if (vlen > UIO_MAXIOV) |
2202 | vlen = UIO_MAXIOV; | |
228e548e AB |
2203 | |
2204 | datagrams = 0; | |
2205 | ||
2206 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2207 | if (!sock) | |
2208 | return err; | |
2209 | ||
c71d8ebe | 2210 | used_address.name_len = UINT_MAX; |
228e548e AB |
2211 | entry = mmsg; |
2212 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2213 | err = 0; |
f092276d | 2214 | flags |= MSG_BATCH; |
228e548e AB |
2215 | |
2216 | while (datagrams < vlen) { | |
f092276d TH |
2217 | if (datagrams == vlen - 1) |
2218 | flags = oflags; | |
2219 | ||
228e548e | 2220 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2221 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2222 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2223 | if (err < 0) |
2224 | break; | |
2225 | err = __put_user(err, &compat_entry->msg_len); | |
2226 | ++compat_entry; | |
2227 | } else { | |
a7526eb5 | 2228 | err = ___sys_sendmsg(sock, |
666547ff | 2229 | (struct user_msghdr __user *)entry, |
28a94d8f | 2230 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2231 | if (err < 0) |
2232 | break; | |
2233 | err = put_user(err, &entry->msg_len); | |
2234 | ++entry; | |
2235 | } | |
2236 | ||
2237 | if (err) | |
2238 | break; | |
2239 | ++datagrams; | |
3023898b SHY |
2240 | if (msg_data_left(&msg_sys)) |
2241 | break; | |
a78cb84c | 2242 | cond_resched(); |
228e548e AB |
2243 | } |
2244 | ||
228e548e AB |
2245 | fput_light(sock->file, fput_needed); |
2246 | ||
728ffb86 AB |
2247 | /* We only return an error if no datagrams were able to be sent */ |
2248 | if (datagrams != 0) | |
228e548e AB |
2249 | return datagrams; |
2250 | ||
228e548e AB |
2251 | return err; |
2252 | } | |
2253 | ||
2254 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2255 | unsigned int, vlen, unsigned int, flags) | |
2256 | { | |
e1834a32 | 2257 | return __sys_sendmmsg(fd, mmsg, vlen, flags, true); |
228e548e AB |
2258 | } |
2259 | ||
666547ff | 2260 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2261 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2262 | { |
89bddce5 SH |
2263 | struct compat_msghdr __user *msg_compat = |
2264 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2265 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2266 | struct iovec *iov = iovstack; |
1da177e4 | 2267 | unsigned long cmsg_ptr; |
2da62906 | 2268 | int len; |
08adb7da | 2269 | ssize_t err; |
1da177e4 LT |
2270 | |
2271 | /* kernel mode address */ | |
230b1839 | 2272 | struct sockaddr_storage addr; |
1da177e4 LT |
2273 | |
2274 | /* user mode address pointers */ | |
2275 | struct sockaddr __user *uaddr; | |
08adb7da | 2276 | int __user *uaddr_len = COMPAT_NAMELEN(msg); |
89bddce5 | 2277 | |
08adb7da | 2278 | msg_sys->msg_name = &addr; |
1da177e4 | 2279 | |
f3d33426 | 2280 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2281 | err = get_compat_msghdr(msg_sys, msg_compat, &uaddr, &iov); |
f3d33426 | 2282 | else |
08adb7da | 2283 | err = copy_msghdr_from_user(msg_sys, msg, &uaddr, &iov); |
1da177e4 | 2284 | if (err < 0) |
da184284 | 2285 | return err; |
1da177e4 | 2286 | |
a2e27255 ACM |
2287 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2288 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2289 | |
f3d33426 HFS |
2290 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2291 | msg_sys->msg_namelen = 0; | |
2292 | ||
1da177e4 LT |
2293 | if (sock->file->f_flags & O_NONBLOCK) |
2294 | flags |= MSG_DONTWAIT; | |
2da62906 | 2295 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, flags); |
1da177e4 LT |
2296 | if (err < 0) |
2297 | goto out_freeiov; | |
2298 | len = err; | |
2299 | ||
2300 | if (uaddr != NULL) { | |
43db362d | 2301 | err = move_addr_to_user(&addr, |
a2e27255 | 2302 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2303 | uaddr_len); |
1da177e4 LT |
2304 | if (err < 0) |
2305 | goto out_freeiov; | |
2306 | } | |
a2e27255 | 2307 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2308 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2309 | if (err) |
2310 | goto out_freeiov; | |
2311 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2312 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2313 | &msg_compat->msg_controllen); |
2314 | else | |
a2e27255 | 2315 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2316 | &msg->msg_controllen); |
2317 | if (err) | |
2318 | goto out_freeiov; | |
2319 | err = len; | |
2320 | ||
2321 | out_freeiov: | |
da184284 | 2322 | kfree(iov); |
a2e27255 ACM |
2323 | return err; |
2324 | } | |
2325 | ||
2326 | /* | |
2327 | * BSD recvmsg interface | |
2328 | */ | |
2329 | ||
e1834a32 DB |
2330 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2331 | bool forbid_cmsg_compat) | |
a2e27255 ACM |
2332 | { |
2333 | int fput_needed, err; | |
2334 | struct msghdr msg_sys; | |
1be374a0 AL |
2335 | struct socket *sock; |
2336 | ||
e1834a32 DB |
2337 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2338 | return -EINVAL; | |
2339 | ||
1be374a0 | 2340 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2341 | if (!sock) |
2342 | goto out; | |
2343 | ||
a7526eb5 | 2344 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2345 | |
6cb153ca | 2346 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2347 | out: |
2348 | return err; | |
2349 | } | |
2350 | ||
666547ff | 2351 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2352 | unsigned int, flags) |
2353 | { | |
e1834a32 | 2354 | return __sys_recvmsg(fd, msg, flags, true); |
a7526eb5 AL |
2355 | } |
2356 | ||
a2e27255 ACM |
2357 | /* |
2358 | * Linux recvmmsg interface | |
2359 | */ | |
2360 | ||
2361 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2362 | unsigned int flags, struct timespec *timeout) | |
2363 | { | |
2364 | int fput_needed, err, datagrams; | |
2365 | struct socket *sock; | |
2366 | struct mmsghdr __user *entry; | |
d7256d0e | 2367 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2368 | struct msghdr msg_sys; |
766b9f92 DD |
2369 | struct timespec64 end_time; |
2370 | struct timespec64 timeout64; | |
a2e27255 ACM |
2371 | |
2372 | if (timeout && | |
2373 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2374 | timeout->tv_nsec)) | |
2375 | return -EINVAL; | |
2376 | ||
2377 | datagrams = 0; | |
2378 | ||
2379 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2380 | if (!sock) | |
2381 | return err; | |
2382 | ||
7797dc41 SHY |
2383 | if (likely(!(flags & MSG_ERRQUEUE))) { |
2384 | err = sock_error(sock->sk); | |
2385 | if (err) { | |
2386 | datagrams = err; | |
2387 | goto out_put; | |
2388 | } | |
e623a9e9 | 2389 | } |
a2e27255 ACM |
2390 | |
2391 | entry = mmsg; | |
d7256d0e | 2392 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2393 | |
2394 | while (datagrams < vlen) { | |
2395 | /* | |
2396 | * No need to ask LSM for more than the first datagram. | |
2397 | */ | |
d7256d0e | 2398 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2399 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2400 | &msg_sys, flags & ~MSG_WAITFORONE, |
2401 | datagrams); | |
d7256d0e JMG |
2402 | if (err < 0) |
2403 | break; | |
2404 | err = __put_user(err, &compat_entry->msg_len); | |
2405 | ++compat_entry; | |
2406 | } else { | |
a7526eb5 | 2407 | err = ___sys_recvmsg(sock, |
666547ff | 2408 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2409 | &msg_sys, flags & ~MSG_WAITFORONE, |
2410 | datagrams); | |
d7256d0e JMG |
2411 | if (err < 0) |
2412 | break; | |
2413 | err = put_user(err, &entry->msg_len); | |
2414 | ++entry; | |
2415 | } | |
2416 | ||
a2e27255 ACM |
2417 | if (err) |
2418 | break; | |
a2e27255 ACM |
2419 | ++datagrams; |
2420 | ||
71c5c159 BB |
2421 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2422 | if (flags & MSG_WAITFORONE) | |
2423 | flags |= MSG_DONTWAIT; | |
2424 | ||
a2e27255 | 2425 | if (timeout) { |
766b9f92 DD |
2426 | ktime_get_ts64(&timeout64); |
2427 | *timeout = timespec64_to_timespec( | |
2428 | timespec64_sub(end_time, timeout64)); | |
a2e27255 ACM |
2429 | if (timeout->tv_sec < 0) { |
2430 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2431 | break; | |
2432 | } | |
2433 | ||
2434 | /* Timeout, return less than vlen datagrams */ | |
2435 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2436 | break; | |
2437 | } | |
2438 | ||
2439 | /* Out of band data, return right away */ | |
2440 | if (msg_sys.msg_flags & MSG_OOB) | |
2441 | break; | |
a78cb84c | 2442 | cond_resched(); |
a2e27255 ACM |
2443 | } |
2444 | ||
a2e27255 | 2445 | if (err == 0) |
34b88a68 ACM |
2446 | goto out_put; |
2447 | ||
2448 | if (datagrams == 0) { | |
2449 | datagrams = err; | |
2450 | goto out_put; | |
2451 | } | |
a2e27255 | 2452 | |
34b88a68 ACM |
2453 | /* |
2454 | * We may return less entries than requested (vlen) if the | |
2455 | * sock is non block and there aren't enough datagrams... | |
2456 | */ | |
2457 | if (err != -EAGAIN) { | |
a2e27255 | 2458 | /* |
34b88a68 ACM |
2459 | * ... or if recvmsg returns an error after we |
2460 | * received some datagrams, where we record the | |
2461 | * error to return on the next call or if the | |
2462 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2463 | */ |
34b88a68 | 2464 | sock->sk->sk_err = -err; |
a2e27255 | 2465 | } |
34b88a68 ACM |
2466 | out_put: |
2467 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2468 | |
34b88a68 | 2469 | return datagrams; |
a2e27255 ACM |
2470 | } |
2471 | ||
1255e269 DB |
2472 | static int do_sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2473 | unsigned int vlen, unsigned int flags, | |
2474 | struct timespec __user *timeout) | |
a2e27255 ACM |
2475 | { |
2476 | int datagrams; | |
2477 | struct timespec timeout_sys; | |
2478 | ||
1be374a0 AL |
2479 | if (flags & MSG_CMSG_COMPAT) |
2480 | return -EINVAL; | |
2481 | ||
a2e27255 ACM |
2482 | if (!timeout) |
2483 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2484 | ||
2485 | if (copy_from_user(&timeout_sys, timeout, sizeof(timeout_sys))) | |
2486 | return -EFAULT; | |
2487 | ||
2488 | datagrams = __sys_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
2489 | ||
2490 | if (datagrams > 0 && | |
2491 | copy_to_user(timeout, &timeout_sys, sizeof(timeout_sys))) | |
2492 | datagrams = -EFAULT; | |
2493 | ||
2494 | return datagrams; | |
2495 | } | |
2496 | ||
1255e269 DB |
2497 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, |
2498 | unsigned int, vlen, unsigned int, flags, | |
2499 | struct timespec __user *, timeout) | |
2500 | { | |
2501 | return do_sys_recvmmsg(fd, mmsg, vlen, flags, timeout); | |
2502 | } | |
2503 | ||
a2e27255 | 2504 | #ifdef __ARCH_WANT_SYS_SOCKETCALL |
1da177e4 LT |
2505 | /* Argument list sizes for sys_socketcall */ |
2506 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2507 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2508 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2509 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2510 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2511 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2512 | }; |
2513 | ||
1da177e4 LT |
2514 | #undef AL |
2515 | ||
2516 | /* | |
89bddce5 | 2517 | * System call vectors. |
1da177e4 LT |
2518 | * |
2519 | * Argument checking cleaned up. Saved 20% in size. | |
2520 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2521 | * it is set by the callees. |
1da177e4 LT |
2522 | */ |
2523 | ||
3e0fa65f | 2524 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2525 | { |
2950fa9d | 2526 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2527 | unsigned long a0, a1; |
1da177e4 | 2528 | int err; |
47379052 | 2529 | unsigned int len; |
1da177e4 | 2530 | |
228e548e | 2531 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 | 2532 | return -EINVAL; |
c8e8cd57 | 2533 | call = array_index_nospec(call, SYS_SENDMMSG + 1); |
1da177e4 | 2534 | |
47379052 AV |
2535 | len = nargs[call]; |
2536 | if (len > sizeof(a)) | |
2537 | return -EINVAL; | |
2538 | ||
1da177e4 | 2539 | /* copy_from_user should be SMP safe. */ |
47379052 | 2540 | if (copy_from_user(a, args, len)) |
1da177e4 | 2541 | return -EFAULT; |
3ec3b2fb | 2542 | |
2950fa9d CG |
2543 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2544 | if (err) | |
2545 | return err; | |
3ec3b2fb | 2546 | |
89bddce5 SH |
2547 | a0 = a[0]; |
2548 | a1 = a[1]; | |
2549 | ||
2550 | switch (call) { | |
2551 | case SYS_SOCKET: | |
9d6a15c3 | 2552 | err = __sys_socket(a0, a1, a[2]); |
89bddce5 SH |
2553 | break; |
2554 | case SYS_BIND: | |
a87d35d8 | 2555 | err = __sys_bind(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2556 | break; |
2557 | case SYS_CONNECT: | |
1387c2c2 | 2558 | err = __sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2559 | break; |
2560 | case SYS_LISTEN: | |
25e290ee | 2561 | err = __sys_listen(a0, a1); |
89bddce5 SH |
2562 | break; |
2563 | case SYS_ACCEPT: | |
4541e805 DB |
2564 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2565 | (int __user *)a[2], 0); | |
89bddce5 SH |
2566 | break; |
2567 | case SYS_GETSOCKNAME: | |
2568 | err = | |
8882a107 DB |
2569 | __sys_getsockname(a0, (struct sockaddr __user *)a1, |
2570 | (int __user *)a[2]); | |
89bddce5 SH |
2571 | break; |
2572 | case SYS_GETPEERNAME: | |
2573 | err = | |
b21c8f83 DB |
2574 | __sys_getpeername(a0, (struct sockaddr __user *)a1, |
2575 | (int __user *)a[2]); | |
89bddce5 SH |
2576 | break; |
2577 | case SYS_SOCKETPAIR: | |
6debc8d8 | 2578 | err = __sys_socketpair(a0, a1, a[2], (int __user *)a[3]); |
89bddce5 SH |
2579 | break; |
2580 | case SYS_SEND: | |
f3bf896b DB |
2581 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2582 | NULL, 0); | |
89bddce5 SH |
2583 | break; |
2584 | case SYS_SENDTO: | |
211b634b DB |
2585 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2586 | (struct sockaddr __user *)a[4], a[5]); | |
89bddce5 SH |
2587 | break; |
2588 | case SYS_RECV: | |
d27e9afc DB |
2589 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2590 | NULL, NULL); | |
89bddce5 SH |
2591 | break; |
2592 | case SYS_RECVFROM: | |
7a09e1eb DB |
2593 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2594 | (struct sockaddr __user *)a[4], | |
2595 | (int __user *)a[5]); | |
89bddce5 SH |
2596 | break; |
2597 | case SYS_SHUTDOWN: | |
005a1aea | 2598 | err = __sys_shutdown(a0, a1); |
89bddce5 SH |
2599 | break; |
2600 | case SYS_SETSOCKOPT: | |
cc36dca0 DB |
2601 | err = __sys_setsockopt(a0, a1, a[2], (char __user *)a[3], |
2602 | a[4]); | |
89bddce5 SH |
2603 | break; |
2604 | case SYS_GETSOCKOPT: | |
2605 | err = | |
13a2d70e DB |
2606 | __sys_getsockopt(a0, a1, a[2], (char __user *)a[3], |
2607 | (int __user *)a[4]); | |
89bddce5 SH |
2608 | break; |
2609 | case SYS_SENDMSG: | |
e1834a32 DB |
2610 | err = __sys_sendmsg(a0, (struct user_msghdr __user *)a1, |
2611 | a[2], true); | |
89bddce5 | 2612 | break; |
228e548e | 2613 | case SYS_SENDMMSG: |
e1834a32 DB |
2614 | err = __sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
2615 | a[3], true); | |
228e548e | 2616 | break; |
89bddce5 | 2617 | case SYS_RECVMSG: |
e1834a32 DB |
2618 | err = __sys_recvmsg(a0, (struct user_msghdr __user *)a1, |
2619 | a[2], true); | |
89bddce5 | 2620 | break; |
a2e27255 | 2621 | case SYS_RECVMMSG: |
1255e269 DB |
2622 | err = do_sys_recvmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
2623 | a[3], (struct timespec __user *)a[4]); | |
a2e27255 | 2624 | break; |
de11defe | 2625 | case SYS_ACCEPT4: |
4541e805 DB |
2626 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2627 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2628 | break; |
89bddce5 SH |
2629 | default: |
2630 | err = -EINVAL; | |
2631 | break; | |
1da177e4 LT |
2632 | } |
2633 | return err; | |
2634 | } | |
2635 | ||
89bddce5 | 2636 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2637 | |
55737fda SH |
2638 | /** |
2639 | * sock_register - add a socket protocol handler | |
2640 | * @ops: description of protocol | |
2641 | * | |
1da177e4 LT |
2642 | * This function is called by a protocol handler that wants to |
2643 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2644 | * socket interface. The value ops->family corresponds to the |
55737fda | 2645 | * socket system call protocol family. |
1da177e4 | 2646 | */ |
f0fd27d4 | 2647 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2648 | { |
2649 | int err; | |
2650 | ||
2651 | if (ops->family >= NPROTO) { | |
3410f22e | 2652 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2653 | return -ENOBUFS; |
2654 | } | |
55737fda SH |
2655 | |
2656 | spin_lock(&net_family_lock); | |
190683a9 ED |
2657 | if (rcu_dereference_protected(net_families[ops->family], |
2658 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2659 | err = -EEXIST; |
2660 | else { | |
cf778b00 | 2661 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2662 | err = 0; |
2663 | } | |
55737fda SH |
2664 | spin_unlock(&net_family_lock); |
2665 | ||
3410f22e | 2666 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2667 | return err; |
2668 | } | |
c6d409cf | 2669 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2670 | |
55737fda SH |
2671 | /** |
2672 | * sock_unregister - remove a protocol handler | |
2673 | * @family: protocol family to remove | |
2674 | * | |
1da177e4 LT |
2675 | * This function is called by a protocol handler that wants to |
2676 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2677 | * new socket creation. |
2678 | * | |
2679 | * If protocol handler is a module, then it can use module reference | |
2680 | * counts to protect against new references. If protocol handler is not | |
2681 | * a module then it needs to provide its own protection in | |
2682 | * the ops->create routine. | |
1da177e4 | 2683 | */ |
f0fd27d4 | 2684 | void sock_unregister(int family) |
1da177e4 | 2685 | { |
f0fd27d4 | 2686 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2687 | |
55737fda | 2688 | spin_lock(&net_family_lock); |
a9b3cd7f | 2689 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2690 | spin_unlock(&net_family_lock); |
2691 | ||
2692 | synchronize_rcu(); | |
2693 | ||
3410f22e | 2694 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2695 | } |
c6d409cf | 2696 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2697 | |
bf2ae2e4 XL |
2698 | bool sock_is_registered(int family) |
2699 | { | |
e978de7a JC |
2700 | return family < NPROTO && |
2701 | rcu_access_pointer(net_families[array_index_nospec(family, NPROTO)]); | |
bf2ae2e4 XL |
2702 | } |
2703 | ||
77d76ea3 | 2704 | static int __init sock_init(void) |
1da177e4 | 2705 | { |
b3e19d92 | 2706 | int err; |
2ca794e5 EB |
2707 | /* |
2708 | * Initialize the network sysctl infrastructure. | |
2709 | */ | |
2710 | err = net_sysctl_init(); | |
2711 | if (err) | |
2712 | goto out; | |
b3e19d92 | 2713 | |
1da177e4 | 2714 | /* |
89bddce5 | 2715 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2716 | */ |
2717 | skb_init(); | |
1da177e4 LT |
2718 | |
2719 | /* | |
89bddce5 | 2720 | * Initialize the protocols module. |
1da177e4 LT |
2721 | */ |
2722 | ||
2723 | init_inodecache(); | |
b3e19d92 NP |
2724 | |
2725 | err = register_filesystem(&sock_fs_type); | |
2726 | if (err) | |
2727 | goto out_fs; | |
1da177e4 | 2728 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2729 | if (IS_ERR(sock_mnt)) { |
2730 | err = PTR_ERR(sock_mnt); | |
2731 | goto out_mount; | |
2732 | } | |
77d76ea3 AK |
2733 | |
2734 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2735 | */ |
2736 | ||
2737 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
2738 | err = netfilter_init(); |
2739 | if (err) | |
2740 | goto out; | |
1da177e4 | 2741 | #endif |
cbeb321a | 2742 | |
408eccce | 2743 | ptp_classifier_init(); |
c1f19b51 | 2744 | |
b3e19d92 NP |
2745 | out: |
2746 | return err; | |
2747 | ||
2748 | out_mount: | |
2749 | unregister_filesystem(&sock_fs_type); | |
2750 | out_fs: | |
2751 | goto out; | |
1da177e4 LT |
2752 | } |
2753 | ||
77d76ea3 AK |
2754 | core_initcall(sock_init); /* early initcall */ |
2755 | ||
1da177e4 LT |
2756 | #ifdef CONFIG_PROC_FS |
2757 | void socket_seq_show(struct seq_file *seq) | |
2758 | { | |
648845ab TZ |
2759 | seq_printf(seq, "sockets: used %d\n", |
2760 | sock_inuse_get(seq->private)); | |
1da177e4 | 2761 | } |
89bddce5 | 2762 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 2763 | |
89bbfc95 | 2764 | #ifdef CONFIG_COMPAT |
6b96018b | 2765 | static int do_siocgstamp(struct net *net, struct socket *sock, |
644595f8 | 2766 | unsigned int cmd, void __user *up) |
7a229387 | 2767 | { |
7a229387 AB |
2768 | mm_segment_t old_fs = get_fs(); |
2769 | struct timeval ktv; | |
2770 | int err; | |
2771 | ||
2772 | set_fs(KERNEL_DS); | |
6b96018b | 2773 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&ktv); |
7a229387 | 2774 | set_fs(old_fs); |
644595f8 | 2775 | if (!err) |
ed6fe9d6 | 2776 | err = compat_put_timeval(&ktv, up); |
644595f8 | 2777 | |
7a229387 AB |
2778 | return err; |
2779 | } | |
2780 | ||
6b96018b | 2781 | static int do_siocgstampns(struct net *net, struct socket *sock, |
644595f8 | 2782 | unsigned int cmd, void __user *up) |
7a229387 | 2783 | { |
7a229387 AB |
2784 | mm_segment_t old_fs = get_fs(); |
2785 | struct timespec kts; | |
2786 | int err; | |
2787 | ||
2788 | set_fs(KERNEL_DS); | |
6b96018b | 2789 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&kts); |
7a229387 | 2790 | set_fs(old_fs); |
644595f8 | 2791 | if (!err) |
ed6fe9d6 | 2792 | err = compat_put_timespec(&kts, up); |
644595f8 | 2793 | |
7a229387 AB |
2794 | return err; |
2795 | } | |
2796 | ||
36fd633e | 2797 | static int compat_dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 2798 | { |
6b96018b | 2799 | struct compat_ifconf ifc32; |
7a229387 | 2800 | struct ifconf ifc; |
7a229387 AB |
2801 | int err; |
2802 | ||
6b96018b | 2803 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2804 | return -EFAULT; |
2805 | ||
36fd633e AV |
2806 | ifc.ifc_len = ifc32.ifc_len; |
2807 | ifc.ifc_req = compat_ptr(ifc32.ifcbuf); | |
7a229387 | 2808 | |
36fd633e AV |
2809 | rtnl_lock(); |
2810 | err = dev_ifconf(net, &ifc, sizeof(struct compat_ifreq)); | |
2811 | rtnl_unlock(); | |
7a229387 AB |
2812 | if (err) |
2813 | return err; | |
2814 | ||
36fd633e | 2815 | ifc32.ifc_len = ifc.ifc_len; |
6b96018b | 2816 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2817 | return -EFAULT; |
2818 | ||
2819 | return 0; | |
2820 | } | |
2821 | ||
6b96018b | 2822 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 2823 | { |
3a7da39d BH |
2824 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
2825 | bool convert_in = false, convert_out = false; | |
44c02a2c AV |
2826 | size_t buf_size = 0; |
2827 | struct ethtool_rxnfc __user *rxnfc = NULL; | |
2828 | struct ifreq ifr; | |
3a7da39d BH |
2829 | u32 rule_cnt = 0, actual_rule_cnt; |
2830 | u32 ethcmd; | |
7a229387 | 2831 | u32 data; |
3a7da39d | 2832 | int ret; |
7a229387 | 2833 | |
3a7da39d BH |
2834 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
2835 | return -EFAULT; | |
7a229387 | 2836 | |
3a7da39d BH |
2837 | compat_rxnfc = compat_ptr(data); |
2838 | ||
2839 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
2840 | return -EFAULT; |
2841 | ||
3a7da39d BH |
2842 | /* Most ethtool structures are defined without padding. |
2843 | * Unfortunately struct ethtool_rxnfc is an exception. | |
2844 | */ | |
2845 | switch (ethcmd) { | |
2846 | default: | |
2847 | break; | |
2848 | case ETHTOOL_GRXCLSRLALL: | |
2849 | /* Buffer size is variable */ | |
2850 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
2851 | return -EFAULT; | |
2852 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
2853 | return -ENOMEM; | |
2854 | buf_size += rule_cnt * sizeof(u32); | |
2855 | /* fall through */ | |
2856 | case ETHTOOL_GRXRINGS: | |
2857 | case ETHTOOL_GRXCLSRLCNT: | |
2858 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 2859 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
2860 | convert_out = true; |
2861 | /* fall through */ | |
2862 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
2863 | buf_size += sizeof(struct ethtool_rxnfc); |
2864 | convert_in = true; | |
44c02a2c | 2865 | rxnfc = compat_alloc_user_space(buf_size); |
3a7da39d BH |
2866 | break; |
2867 | } | |
2868 | ||
44c02a2c | 2869 | if (copy_from_user(&ifr.ifr_name, &ifr32->ifr_name, IFNAMSIZ)) |
7a229387 AB |
2870 | return -EFAULT; |
2871 | ||
44c02a2c | 2872 | ifr.ifr_data = convert_in ? rxnfc : (void __user *)compat_rxnfc; |
7a229387 | 2873 | |
3a7da39d | 2874 | if (convert_in) { |
127fe533 | 2875 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
2876 | * fs.ring_cookie and at the end of fs, but nowhere else. |
2877 | */ | |
127fe533 AD |
2878 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
2879 | sizeof(compat_rxnfc->fs.m_ext) != | |
2880 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
2881 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
2882 | BUILD_BUG_ON( |
2883 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
2884 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
2885 | offsetof(struct ethtool_rxnfc, fs.location) - | |
2886 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
2887 | ||
2888 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
2889 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
2890 | (void __user *)rxnfc) || | |
3a7da39d BH |
2891 | copy_in_user(&rxnfc->fs.ring_cookie, |
2892 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 SH |
2893 | (void __user *)(&rxnfc->fs.location + 1) - |
2894 | (void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2895 | copy_in_user(&rxnfc->rule_cnt, &compat_rxnfc->rule_cnt, |
2896 | sizeof(rxnfc->rule_cnt))) | |
2897 | return -EFAULT; | |
2898 | } | |
2899 | ||
44c02a2c | 2900 | ret = dev_ioctl(net, SIOCETHTOOL, &ifr, NULL); |
3a7da39d BH |
2901 | if (ret) |
2902 | return ret; | |
2903 | ||
2904 | if (convert_out) { | |
2905 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
2906 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
2907 | (const void __user *)rxnfc) || | |
3a7da39d BH |
2908 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
2909 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
2910 | (const void __user *)(&rxnfc->fs.location + 1) - |
2911 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2912 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
2913 | sizeof(rxnfc->rule_cnt))) | |
2914 | return -EFAULT; | |
2915 | ||
2916 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
2917 | /* As an optimisation, we only copy the actual | |
2918 | * number of rules that the underlying | |
2919 | * function returned. Since Mallory might | |
2920 | * change the rule count in user memory, we | |
2921 | * check that it is less than the rule count | |
2922 | * originally given (as the user buffer size), | |
2923 | * which has been range-checked. | |
2924 | */ | |
2925 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
2926 | return -EFAULT; | |
2927 | if (actual_rule_cnt < rule_cnt) | |
2928 | rule_cnt = actual_rule_cnt; | |
2929 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
2930 | &rxnfc->rule_locs[0], | |
2931 | rule_cnt * sizeof(u32))) | |
2932 | return -EFAULT; | |
2933 | } | |
2934 | } | |
2935 | ||
2936 | return 0; | |
7a229387 AB |
2937 | } |
2938 | ||
7a50a240 AB |
2939 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
2940 | { | |
7a50a240 | 2941 | compat_uptr_t uptr32; |
44c02a2c AV |
2942 | struct ifreq ifr; |
2943 | void __user *saved; | |
2944 | int err; | |
7a50a240 | 2945 | |
44c02a2c | 2946 | if (copy_from_user(&ifr, uifr32, sizeof(struct compat_ifreq))) |
7a50a240 AB |
2947 | return -EFAULT; |
2948 | ||
2949 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
2950 | return -EFAULT; | |
2951 | ||
44c02a2c AV |
2952 | saved = ifr.ifr_settings.ifs_ifsu.raw_hdlc; |
2953 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = compat_ptr(uptr32); | |
7a229387 | 2954 | |
44c02a2c AV |
2955 | err = dev_ioctl(net, SIOCWANDEV, &ifr, NULL); |
2956 | if (!err) { | |
2957 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = saved; | |
2958 | if (copy_to_user(uifr32, &ifr, sizeof(struct compat_ifreq))) | |
2959 | err = -EFAULT; | |
ccbd6a5a | 2960 | } |
44c02a2c | 2961 | return err; |
7a229387 AB |
2962 | } |
2963 | ||
590d4693 BH |
2964 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
2965 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 2966 | struct compat_ifreq __user *u_ifreq32) |
7a229387 | 2967 | { |
44c02a2c | 2968 | struct ifreq ifreq; |
7a229387 AB |
2969 | u32 data32; |
2970 | ||
44c02a2c | 2971 | if (copy_from_user(ifreq.ifr_name, u_ifreq32->ifr_name, IFNAMSIZ)) |
7a229387 | 2972 | return -EFAULT; |
44c02a2c | 2973 | if (get_user(data32, &u_ifreq32->ifr_data)) |
7a229387 | 2974 | return -EFAULT; |
44c02a2c | 2975 | ifreq.ifr_data = compat_ptr(data32); |
7a229387 | 2976 | |
44c02a2c | 2977 | return dev_ioctl(net, cmd, &ifreq, NULL); |
7a229387 AB |
2978 | } |
2979 | ||
a2116ed2 AB |
2980 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
2981 | struct compat_ifreq __user *uifr32) | |
2982 | { | |
2983 | struct ifreq ifr; | |
2984 | struct compat_ifmap __user *uifmap32; | |
a2116ed2 AB |
2985 | int err; |
2986 | ||
2987 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
2988 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
2989 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
2990 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
2991 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
2992 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
2993 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
2994 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
2995 | if (err) |
2996 | return -EFAULT; | |
2997 | ||
44c02a2c | 2998 | err = dev_ioctl(net, cmd, &ifr, NULL); |
a2116ed2 AB |
2999 | |
3000 | if (cmd == SIOCGIFMAP && !err) { | |
3001 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3002 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3003 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3004 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3005 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3006 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3007 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3008 | if (err) |
3009 | err = -EFAULT; | |
3010 | } | |
3011 | return err; | |
3012 | } | |
3013 | ||
7a229387 | 3014 | struct rtentry32 { |
c6d409cf | 3015 | u32 rt_pad1; |
7a229387 AB |
3016 | struct sockaddr rt_dst; /* target address */ |
3017 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
3018 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
3019 | unsigned short rt_flags; |
3020 | short rt_pad2; | |
3021 | u32 rt_pad3; | |
3022 | unsigned char rt_tos; | |
3023 | unsigned char rt_class; | |
3024 | short rt_pad4; | |
3025 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 3026 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
3027 | u32 rt_mtu; /* per route MTU/Window */ |
3028 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
3029 | unsigned short rt_irtt; /* Initial RTT */ |
3030 | }; | |
3031 | ||
3032 | struct in6_rtmsg32 { | |
3033 | struct in6_addr rtmsg_dst; | |
3034 | struct in6_addr rtmsg_src; | |
3035 | struct in6_addr rtmsg_gateway; | |
3036 | u32 rtmsg_type; | |
3037 | u16 rtmsg_dst_len; | |
3038 | u16 rtmsg_src_len; | |
3039 | u32 rtmsg_metric; | |
3040 | u32 rtmsg_info; | |
3041 | u32 rtmsg_flags; | |
3042 | s32 rtmsg_ifindex; | |
3043 | }; | |
3044 | ||
6b96018b AB |
3045 | static int routing_ioctl(struct net *net, struct socket *sock, |
3046 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3047 | { |
3048 | int ret; | |
3049 | void *r = NULL; | |
3050 | struct in6_rtmsg r6; | |
3051 | struct rtentry r4; | |
3052 | char devname[16]; | |
3053 | u32 rtdev; | |
3054 | mm_segment_t old_fs = get_fs(); | |
3055 | ||
6b96018b AB |
3056 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3057 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3058 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3059 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
3060 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3061 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3062 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3063 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3064 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3065 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3066 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3067 | |
3068 | r = (void *) &r6; | |
3069 | } else { /* ipv4 */ | |
6b96018b | 3070 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3071 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3072 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
3073 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
3074 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3075 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3076 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
3077 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3078 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3079 | if (rtdev) { |
c6d409cf | 3080 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3081 | r4.rt_dev = (char __user __force *)devname; |
3082 | devname[15] = 0; | |
7a229387 AB |
3083 | } else |
3084 | r4.rt_dev = NULL; | |
3085 | ||
3086 | r = (void *) &r4; | |
3087 | } | |
3088 | ||
3089 | if (ret) { | |
3090 | ret = -EFAULT; | |
3091 | goto out; | |
3092 | } | |
3093 | ||
c6d409cf | 3094 | set_fs(KERNEL_DS); |
6b96018b | 3095 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3096 | set_fs(old_fs); |
7a229387 AB |
3097 | |
3098 | out: | |
7a229387 AB |
3099 | return ret; |
3100 | } | |
3101 | ||
3102 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3103 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3104 | * use compatible ioctls |
7a229387 | 3105 | */ |
6b96018b | 3106 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3107 | { |
6b96018b | 3108 | compat_ulong_t tmp; |
7a229387 | 3109 | |
6b96018b | 3110 | if (get_user(tmp, argp)) |
7a229387 AB |
3111 | return -EFAULT; |
3112 | if (tmp == BRCTL_GET_VERSION) | |
3113 | return BRCTL_VERSION + 1; | |
3114 | return -EINVAL; | |
3115 | } | |
3116 | ||
6b96018b AB |
3117 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3118 | unsigned int cmd, unsigned long arg) | |
3119 | { | |
3120 | void __user *argp = compat_ptr(arg); | |
3121 | struct sock *sk = sock->sk; | |
3122 | struct net *net = sock_net(sk); | |
7a229387 | 3123 | |
6b96018b | 3124 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3125 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3126 | |
3127 | switch (cmd) { | |
3128 | case SIOCSIFBR: | |
3129 | case SIOCGIFBR: | |
3130 | return old_bridge_ioctl(argp); | |
6b96018b | 3131 | case SIOCGIFCONF: |
36fd633e | 3132 | return compat_dev_ifconf(net, argp); |
6b96018b AB |
3133 | case SIOCETHTOOL: |
3134 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3135 | case SIOCWANDEV: |
3136 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3137 | case SIOCGIFMAP: |
3138 | case SIOCSIFMAP: | |
3139 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3140 | case SIOCADDRT: |
3141 | case SIOCDELRT: | |
3142 | return routing_ioctl(net, sock, cmd, argp); | |
3143 | case SIOCGSTAMP: | |
3144 | return do_siocgstamp(net, sock, cmd, argp); | |
3145 | case SIOCGSTAMPNS: | |
3146 | return do_siocgstampns(net, sock, cmd, argp); | |
590d4693 BH |
3147 | case SIOCBONDSLAVEINFOQUERY: |
3148 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3149 | case SIOCSHWTSTAMP: |
fd468c74 | 3150 | case SIOCGHWTSTAMP: |
590d4693 | 3151 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3152 | |
3153 | case FIOSETOWN: | |
3154 | case SIOCSPGRP: | |
3155 | case FIOGETOWN: | |
3156 | case SIOCGPGRP: | |
3157 | case SIOCBRADDBR: | |
3158 | case SIOCBRDELBR: | |
3159 | case SIOCGIFVLAN: | |
3160 | case SIOCSIFVLAN: | |
3161 | case SIOCADDDLCI: | |
3162 | case SIOCDELDLCI: | |
c62cce2c | 3163 | case SIOCGSKNS: |
6b96018b AB |
3164 | return sock_ioctl(file, cmd, arg); |
3165 | ||
3166 | case SIOCGIFFLAGS: | |
3167 | case SIOCSIFFLAGS: | |
3168 | case SIOCGIFMETRIC: | |
3169 | case SIOCSIFMETRIC: | |
3170 | case SIOCGIFMTU: | |
3171 | case SIOCSIFMTU: | |
3172 | case SIOCGIFMEM: | |
3173 | case SIOCSIFMEM: | |
3174 | case SIOCGIFHWADDR: | |
3175 | case SIOCSIFHWADDR: | |
3176 | case SIOCADDMULTI: | |
3177 | case SIOCDELMULTI: | |
3178 | case SIOCGIFINDEX: | |
6b96018b AB |
3179 | case SIOCGIFADDR: |
3180 | case SIOCSIFADDR: | |
3181 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3182 | case SIOCDIFADDR: |
6b96018b AB |
3183 | case SIOCGIFBRDADDR: |
3184 | case SIOCSIFBRDADDR: | |
3185 | case SIOCGIFDSTADDR: | |
3186 | case SIOCSIFDSTADDR: | |
3187 | case SIOCGIFNETMASK: | |
3188 | case SIOCSIFNETMASK: | |
3189 | case SIOCSIFPFLAGS: | |
3190 | case SIOCGIFPFLAGS: | |
3191 | case SIOCGIFTXQLEN: | |
3192 | case SIOCSIFTXQLEN: | |
3193 | case SIOCBRADDIF: | |
3194 | case SIOCBRDELIF: | |
9177efd3 AB |
3195 | case SIOCSIFNAME: |
3196 | case SIOCGMIIPHY: | |
3197 | case SIOCGMIIREG: | |
3198 | case SIOCSMIIREG: | |
6b96018b AB |
3199 | case SIOCSARP: |
3200 | case SIOCGARP: | |
3201 | case SIOCDARP: | |
6b96018b | 3202 | case SIOCATMARK: |
f92d4fc9 AV |
3203 | case SIOCBONDENSLAVE: |
3204 | case SIOCBONDRELEASE: | |
3205 | case SIOCBONDSETHWADDR: | |
3206 | case SIOCBONDCHANGEACTIVE: | |
4cf808e7 | 3207 | case SIOCGIFNAME: |
9177efd3 AB |
3208 | return sock_do_ioctl(net, sock, cmd, arg); |
3209 | } | |
3210 | ||
6b96018b AB |
3211 | return -ENOIOCTLCMD; |
3212 | } | |
7a229387 | 3213 | |
95c96174 | 3214 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3215 | unsigned long arg) |
89bbfc95 SP |
3216 | { |
3217 | struct socket *sock = file->private_data; | |
3218 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3219 | struct sock *sk; |
3220 | struct net *net; | |
3221 | ||
3222 | sk = sock->sk; | |
3223 | net = sock_net(sk); | |
89bbfc95 SP |
3224 | |
3225 | if (sock->ops->compat_ioctl) | |
3226 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3227 | ||
87de87d5 DM |
3228 | if (ret == -ENOIOCTLCMD && |
3229 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3230 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3231 | ||
6b96018b AB |
3232 | if (ret == -ENOIOCTLCMD) |
3233 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3234 | ||
89bbfc95 SP |
3235 | return ret; |
3236 | } | |
3237 | #endif | |
3238 | ||
ac5a488e SS |
3239 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3240 | { | |
3241 | return sock->ops->bind(sock, addr, addrlen); | |
3242 | } | |
c6d409cf | 3243 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e SS |
3244 | |
3245 | int kernel_listen(struct socket *sock, int backlog) | |
3246 | { | |
3247 | return sock->ops->listen(sock, backlog); | |
3248 | } | |
c6d409cf | 3249 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e SS |
3250 | |
3251 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) | |
3252 | { | |
3253 | struct sock *sk = sock->sk; | |
3254 | int err; | |
3255 | ||
3256 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3257 | newsock); | |
3258 | if (err < 0) | |
3259 | goto done; | |
3260 | ||
cdfbabfb | 3261 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3262 | if (err < 0) { |
3263 | sock_release(*newsock); | |
fa8705b0 | 3264 | *newsock = NULL; |
ac5a488e SS |
3265 | goto done; |
3266 | } | |
3267 | ||
3268 | (*newsock)->ops = sock->ops; | |
1b08534e | 3269 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3270 | |
3271 | done: | |
3272 | return err; | |
3273 | } | |
c6d409cf | 3274 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e SS |
3275 | |
3276 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, | |
4768fbcb | 3277 | int flags) |
ac5a488e SS |
3278 | { |
3279 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3280 | } | |
c6d409cf | 3281 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e | 3282 | |
9b2c45d4 | 3283 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3284 | { |
9b2c45d4 | 3285 | return sock->ops->getname(sock, addr, 0); |
ac5a488e | 3286 | } |
c6d409cf | 3287 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e | 3288 | |
9b2c45d4 | 3289 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3290 | { |
9b2c45d4 | 3291 | return sock->ops->getname(sock, addr, 1); |
ac5a488e | 3292 | } |
c6d409cf | 3293 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e SS |
3294 | |
3295 | int kernel_getsockopt(struct socket *sock, int level, int optname, | |
3296 | char *optval, int *optlen) | |
3297 | { | |
3298 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3299 | char __user *uoptval; |
3300 | int __user *uoptlen; | |
ac5a488e SS |
3301 | int err; |
3302 | ||
fb8621bb NK |
3303 | uoptval = (char __user __force *) optval; |
3304 | uoptlen = (int __user __force *) optlen; | |
3305 | ||
ac5a488e SS |
3306 | set_fs(KERNEL_DS); |
3307 | if (level == SOL_SOCKET) | |
fb8621bb | 3308 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3309 | else |
fb8621bb NK |
3310 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3311 | uoptlen); | |
ac5a488e SS |
3312 | set_fs(oldfs); |
3313 | return err; | |
3314 | } | |
c6d409cf | 3315 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e SS |
3316 | |
3317 | int kernel_setsockopt(struct socket *sock, int level, int optname, | |
b7058842 | 3318 | char *optval, unsigned int optlen) |
ac5a488e SS |
3319 | { |
3320 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3321 | char __user *uoptval; |
ac5a488e SS |
3322 | int err; |
3323 | ||
fb8621bb NK |
3324 | uoptval = (char __user __force *) optval; |
3325 | ||
ac5a488e SS |
3326 | set_fs(KERNEL_DS); |
3327 | if (level == SOL_SOCKET) | |
fb8621bb | 3328 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3329 | else |
fb8621bb | 3330 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3331 | optlen); |
3332 | set_fs(oldfs); | |
3333 | return err; | |
3334 | } | |
c6d409cf | 3335 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e SS |
3336 | |
3337 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, | |
3338 | size_t size, int flags) | |
3339 | { | |
3340 | if (sock->ops->sendpage) | |
3341 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3342 | ||
3343 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3344 | } | |
c6d409cf | 3345 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3346 | |
306b13eb TH |
3347 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3348 | size_t size, int flags) | |
3349 | { | |
3350 | struct socket *sock = sk->sk_socket; | |
3351 | ||
3352 | if (sock->ops->sendpage_locked) | |
3353 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3354 | flags); | |
3355 | ||
3356 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3357 | } | |
3358 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3359 | ||
91cf45f0 TM |
3360 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3361 | { | |
3362 | return sock->ops->shutdown(sock, how); | |
3363 | } | |
91cf45f0 | 3364 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 P |
3365 | |
3366 | /* This routine returns the IP overhead imposed by a socket i.e. | |
3367 | * the length of the underlying IP header, depending on whether | |
3368 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
57240d00 | 3369 | * on at the socket. Assumes that the caller has a lock on the socket. |
113c3075 P |
3370 | */ |
3371 | u32 kernel_sock_ip_overhead(struct sock *sk) | |
3372 | { | |
3373 | struct inet_sock *inet; | |
3374 | struct ip_options_rcu *opt; | |
3375 | u32 overhead = 0; | |
113c3075 P |
3376 | #if IS_ENABLED(CONFIG_IPV6) |
3377 | struct ipv6_pinfo *np; | |
3378 | struct ipv6_txoptions *optv6 = NULL; | |
3379 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3380 | ||
3381 | if (!sk) | |
3382 | return overhead; | |
3383 | ||
113c3075 P |
3384 | switch (sk->sk_family) { |
3385 | case AF_INET: | |
3386 | inet = inet_sk(sk); | |
3387 | overhead += sizeof(struct iphdr); | |
3388 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3389 | sock_owned_by_user(sk)); |
113c3075 P |
3390 | if (opt) |
3391 | overhead += opt->opt.optlen; | |
3392 | return overhead; | |
3393 | #if IS_ENABLED(CONFIG_IPV6) | |
3394 | case AF_INET6: | |
3395 | np = inet6_sk(sk); | |
3396 | overhead += sizeof(struct ipv6hdr); | |
3397 | if (np) | |
3398 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3399 | sock_owned_by_user(sk)); |
113c3075 P |
3400 | if (optv6) |
3401 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3402 | return overhead; | |
3403 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3404 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3405 | return overhead; | |
3406 | } | |
3407 | } | |
3408 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |