Commit | Line | Data |
---|---|---|
2874c5fd | 1 | // SPDX-License-Identifier: GPL-2.0-or-later |
1da177e4 LT |
2 | /* |
3 | * NET An implementation of the SOCKET network access protocol. | |
4 | * | |
5 | * Version: @(#)socket.c 1.1.93 18/02/95 | |
6 | * | |
7 | * Authors: Orest Zborowski, <obz@Kodak.COM> | |
02c30a84 | 8 | * Ross Biro |
1da177e4 LT |
9 | * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> |
10 | * | |
11 | * Fixes: | |
12 | * Anonymous : NOTSOCK/BADF cleanup. Error fix in | |
13 | * shutdown() | |
14 | * Alan Cox : verify_area() fixes | |
15 | * Alan Cox : Removed DDI | |
16 | * Jonathan Kamens : SOCK_DGRAM reconnect bug | |
17 | * Alan Cox : Moved a load of checks to the very | |
18 | * top level. | |
19 | * Alan Cox : Move address structures to/from user | |
20 | * mode above the protocol layers. | |
21 | * Rob Janssen : Allow 0 length sends. | |
22 | * Alan Cox : Asynchronous I/O support (cribbed from the | |
23 | * tty drivers). | |
24 | * Niibe Yutaka : Asynchronous I/O for writes (4.4BSD style) | |
25 | * Jeff Uphoff : Made max number of sockets command-line | |
26 | * configurable. | |
27 | * Matti Aarnio : Made the number of sockets dynamic, | |
28 | * to be allocated when needed, and mr. | |
29 | * Uphoff's max is used as max to be | |
30 | * allowed to allocate. | |
31 | * Linus : Argh. removed all the socket allocation | |
32 | * altogether: it's in the inode now. | |
33 | * Alan Cox : Made sock_alloc()/sock_release() public | |
34 | * for NetROM and future kernel nfsd type | |
35 | * stuff. | |
36 | * Alan Cox : sendmsg/recvmsg basics. | |
37 | * Tom Dyas : Export net symbols. | |
38 | * Marcin Dalecki : Fixed problems with CONFIG_NET="n". | |
39 | * Alan Cox : Added thread locking to sys_* calls | |
40 | * for sockets. May have errors at the | |
41 | * moment. | |
42 | * Kevin Buhr : Fixed the dumb errors in the above. | |
43 | * Andi Kleen : Some small cleanups, optimizations, | |
44 | * and fixed a copy_from_user() bug. | |
45 | * Tigran Aivazian : sys_send(args) calls sys_sendto(args, NULL, 0) | |
89bddce5 | 46 | * Tigran Aivazian : Made listen(2) backlog sanity checks |
1da177e4 LT |
47 | * protocol-independent |
48 | * | |
1da177e4 | 49 | * This module is effectively the top level interface to the BSD socket |
89bddce5 | 50 | * paradigm. |
1da177e4 LT |
51 | * |
52 | * Based upon Swansea University Computer Society NET3.039 | |
53 | */ | |
54 | ||
cc69837f | 55 | #include <linux/ethtool.h> |
1da177e4 | 56 | #include <linux/mm.h> |
1da177e4 LT |
57 | #include <linux/socket.h> |
58 | #include <linux/file.h> | |
59 | #include <linux/net.h> | |
60 | #include <linux/interrupt.h> | |
aaca0bdc | 61 | #include <linux/thread_info.h> |
55737fda | 62 | #include <linux/rcupdate.h> |
1da177e4 LT |
63 | #include <linux/netdevice.h> |
64 | #include <linux/proc_fs.h> | |
65 | #include <linux/seq_file.h> | |
4a3e2f71 | 66 | #include <linux/mutex.h> |
1da177e4 | 67 | #include <linux/if_bridge.h> |
20380731 | 68 | #include <linux/if_vlan.h> |
408eccce | 69 | #include <linux/ptp_classify.h> |
1da177e4 LT |
70 | #include <linux/init.h> |
71 | #include <linux/poll.h> | |
72 | #include <linux/cache.h> | |
73 | #include <linux/module.h> | |
74 | #include <linux/highmem.h> | |
1da177e4 | 75 | #include <linux/mount.h> |
fba9be49 | 76 | #include <linux/pseudo_fs.h> |
1da177e4 LT |
77 | #include <linux/security.h> |
78 | #include <linux/syscalls.h> | |
79 | #include <linux/compat.h> | |
80 | #include <linux/kmod.h> | |
3ec3b2fb | 81 | #include <linux/audit.h> |
d86b5e0e | 82 | #include <linux/wireless.h> |
1b8d7ae4 | 83 | #include <linux/nsproxy.h> |
1fd7317d | 84 | #include <linux/magic.h> |
5a0e3ad6 | 85 | #include <linux/slab.h> |
600e1779 | 86 | #include <linux/xattr.h> |
c8e8cd57 | 87 | #include <linux/nospec.h> |
8c3c447b | 88 | #include <linux/indirect_call_wrapper.h> |
1da177e4 | 89 | |
7c0f6ba6 | 90 | #include <linux/uaccess.h> |
1da177e4 LT |
91 | #include <asm/unistd.h> |
92 | ||
93 | #include <net/compat.h> | |
87de87d5 | 94 | #include <net/wext.h> |
f8451725 | 95 | #include <net/cls_cgroup.h> |
1da177e4 LT |
96 | |
97 | #include <net/sock.h> | |
98 | #include <linux/netfilter.h> | |
99 | ||
6b96018b AB |
100 | #include <linux/if_tun.h> |
101 | #include <linux/ipv6_route.h> | |
102 | #include <linux/route.h> | |
c7dc504e | 103 | #include <linux/termios.h> |
6b96018b | 104 | #include <linux/sockios.h> |
076bb0c8 | 105 | #include <net/busy_poll.h> |
f24b9be5 | 106 | #include <linux/errqueue.h> |
06021292 | 107 | |
e0d1095a | 108 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
109 | unsigned int sysctl_net_busy_read __read_mostly; |
110 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 111 | #endif |
6b96018b | 112 | |
8ae5e030 AV |
113 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
114 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 115 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
116 | |
117 | static int sock_close(struct inode *inode, struct file *file); | |
a11e1d43 LT |
118 | static __poll_t sock_poll(struct file *file, |
119 | struct poll_table_struct *wait); | |
89bddce5 | 120 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
121 | #ifdef CONFIG_COMPAT |
122 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 123 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 124 | #endif |
1da177e4 | 125 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
126 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
127 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 128 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 129 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 130 | unsigned int flags); |
542d3065 AB |
131 | |
132 | #ifdef CONFIG_PROC_FS | |
133 | static void sock_show_fdinfo(struct seq_file *m, struct file *f) | |
134 | { | |
135 | struct socket *sock = f->private_data; | |
136 | ||
137 | if (sock->ops->show_fdinfo) | |
138 | sock->ops->show_fdinfo(m, sock); | |
139 | } | |
140 | #else | |
141 | #define sock_show_fdinfo NULL | |
142 | #endif | |
1da177e4 | 143 | |
1da177e4 LT |
144 | /* |
145 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
146 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
147 | */ | |
148 | ||
da7071d7 | 149 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
150 | .owner = THIS_MODULE, |
151 | .llseek = no_llseek, | |
8ae5e030 AV |
152 | .read_iter = sock_read_iter, |
153 | .write_iter = sock_write_iter, | |
1da177e4 LT |
154 | .poll = sock_poll, |
155 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
156 | #ifdef CONFIG_COMPAT |
157 | .compat_ioctl = compat_sock_ioctl, | |
158 | #endif | |
1da177e4 | 159 | .mmap = sock_mmap, |
1da177e4 LT |
160 | .release = sock_close, |
161 | .fasync = sock_fasync, | |
5274f052 JA |
162 | .sendpage = sock_sendpage, |
163 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 164 | .splice_read = sock_splice_read, |
b4653342 | 165 | .show_fdinfo = sock_show_fdinfo, |
1da177e4 LT |
166 | }; |
167 | ||
168 | /* | |
169 | * The protocol list. Each protocol is registered in here. | |
170 | */ | |
171 | ||
1da177e4 | 172 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 173 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 174 | |
1da177e4 | 175 | /* |
89bddce5 SH |
176 | * Support routines. |
177 | * Move socket addresses back and forth across the kernel/user | |
178 | * divide and look after the messy bits. | |
1da177e4 LT |
179 | */ |
180 | ||
1da177e4 LT |
181 | /** |
182 | * move_addr_to_kernel - copy a socket address into kernel space | |
183 | * @uaddr: Address in user space | |
184 | * @kaddr: Address in kernel space | |
185 | * @ulen: Length in user space | |
186 | * | |
187 | * The address is copied into kernel space. If the provided address is | |
188 | * too long an error code of -EINVAL is returned. If the copy gives | |
189 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
190 | */ | |
191 | ||
43db362d | 192 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 193 | { |
230b1839 | 194 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 195 | return -EINVAL; |
89bddce5 | 196 | if (ulen == 0) |
1da177e4 | 197 | return 0; |
89bddce5 | 198 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 199 | return -EFAULT; |
3ec3b2fb | 200 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
201 | } |
202 | ||
203 | /** | |
204 | * move_addr_to_user - copy an address to user space | |
205 | * @kaddr: kernel space address | |
206 | * @klen: length of address in kernel | |
207 | * @uaddr: user space address | |
208 | * @ulen: pointer to user length field | |
209 | * | |
210 | * The value pointed to by ulen on entry is the buffer length available. | |
211 | * This is overwritten with the buffer space used. -EINVAL is returned | |
212 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
213 | * is returned if either the buffer or the length field are not | |
214 | * accessible. | |
215 | * After copying the data up to the limit the user specifies, the true | |
216 | * length of the data is written over the length limit the user | |
217 | * specified. Zero is returned for a success. | |
218 | */ | |
89bddce5 | 219 | |
43db362d | 220 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 221 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
222 | { |
223 | int err; | |
224 | int len; | |
225 | ||
68c6beb3 | 226 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
227 | err = get_user(len, ulen); |
228 | if (err) | |
1da177e4 | 229 | return err; |
89bddce5 SH |
230 | if (len > klen) |
231 | len = klen; | |
68c6beb3 | 232 | if (len < 0) |
1da177e4 | 233 | return -EINVAL; |
89bddce5 | 234 | if (len) { |
d6fe3945 SG |
235 | if (audit_sockaddr(klen, kaddr)) |
236 | return -ENOMEM; | |
89bddce5 | 237 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
238 | return -EFAULT; |
239 | } | |
240 | /* | |
89bddce5 SH |
241 | * "fromlen shall refer to the value before truncation.." |
242 | * 1003.1g | |
1da177e4 LT |
243 | */ |
244 | return __put_user(klen, ulen); | |
245 | } | |
246 | ||
08009a76 | 247 | static struct kmem_cache *sock_inode_cachep __ro_after_init; |
1da177e4 LT |
248 | |
249 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
250 | { | |
251 | struct socket_alloc *ei; | |
89bddce5 | 252 | |
e94b1766 | 253 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
254 | if (!ei) |
255 | return NULL; | |
333f7909 AV |
256 | init_waitqueue_head(&ei->socket.wq.wait); |
257 | ei->socket.wq.fasync_list = NULL; | |
258 | ei->socket.wq.flags = 0; | |
89bddce5 | 259 | |
1da177e4 LT |
260 | ei->socket.state = SS_UNCONNECTED; |
261 | ei->socket.flags = 0; | |
262 | ei->socket.ops = NULL; | |
263 | ei->socket.sk = NULL; | |
264 | ei->socket.file = NULL; | |
1da177e4 LT |
265 | |
266 | return &ei->vfs_inode; | |
267 | } | |
268 | ||
6d7855c5 | 269 | static void sock_free_inode(struct inode *inode) |
1da177e4 | 270 | { |
43815482 ED |
271 | struct socket_alloc *ei; |
272 | ||
273 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
43815482 | 274 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
275 | } |
276 | ||
51cc5068 | 277 | static void init_once(void *foo) |
1da177e4 | 278 | { |
89bddce5 | 279 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 280 | |
a35afb83 | 281 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 282 | } |
89bddce5 | 283 | |
1e911632 | 284 | static void init_inodecache(void) |
1da177e4 LT |
285 | { |
286 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
287 | sizeof(struct socket_alloc), |
288 | 0, | |
289 | (SLAB_HWCACHE_ALIGN | | |
290 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 291 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 292 | init_once); |
1e911632 | 293 | BUG_ON(sock_inode_cachep == NULL); |
1da177e4 LT |
294 | } |
295 | ||
b87221de | 296 | static const struct super_operations sockfs_ops = { |
c6d409cf | 297 | .alloc_inode = sock_alloc_inode, |
6d7855c5 | 298 | .free_inode = sock_free_inode, |
c6d409cf | 299 | .statfs = simple_statfs, |
1da177e4 LT |
300 | }; |
301 | ||
c23fbb6b ED |
302 | /* |
303 | * sockfs_dname() is called from d_path(). | |
304 | */ | |
305 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
306 | { | |
307 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 308 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
309 | } |
310 | ||
3ba13d17 | 311 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 312 | .d_dname = sockfs_dname, |
1da177e4 LT |
313 | }; |
314 | ||
bba0bd31 AG |
315 | static int sockfs_xattr_get(const struct xattr_handler *handler, |
316 | struct dentry *dentry, struct inode *inode, | |
317 | const char *suffix, void *value, size_t size) | |
318 | { | |
319 | if (value) { | |
320 | if (dentry->d_name.len + 1 > size) | |
321 | return -ERANGE; | |
322 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
323 | } | |
324 | return dentry->d_name.len + 1; | |
325 | } | |
326 | ||
327 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" | |
328 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
329 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
330 | ||
331 | static const struct xattr_handler sockfs_xattr_handler = { | |
332 | .name = XATTR_NAME_SOCKPROTONAME, | |
333 | .get = sockfs_xattr_get, | |
334 | }; | |
335 | ||
4a590153 | 336 | static int sockfs_security_xattr_set(const struct xattr_handler *handler, |
e65ce2a5 | 337 | struct user_namespace *mnt_userns, |
4a590153 AG |
338 | struct dentry *dentry, struct inode *inode, |
339 | const char *suffix, const void *value, | |
340 | size_t size, int flags) | |
341 | { | |
342 | /* Handled by LSM. */ | |
343 | return -EAGAIN; | |
344 | } | |
345 | ||
346 | static const struct xattr_handler sockfs_security_xattr_handler = { | |
347 | .prefix = XATTR_SECURITY_PREFIX, | |
348 | .set = sockfs_security_xattr_set, | |
349 | }; | |
350 | ||
bba0bd31 AG |
351 | static const struct xattr_handler *sockfs_xattr_handlers[] = { |
352 | &sockfs_xattr_handler, | |
4a590153 | 353 | &sockfs_security_xattr_handler, |
bba0bd31 AG |
354 | NULL |
355 | }; | |
356 | ||
fba9be49 | 357 | static int sockfs_init_fs_context(struct fs_context *fc) |
c74a1cbb | 358 | { |
fba9be49 DH |
359 | struct pseudo_fs_context *ctx = init_pseudo(fc, SOCKFS_MAGIC); |
360 | if (!ctx) | |
361 | return -ENOMEM; | |
362 | ctx->ops = &sockfs_ops; | |
363 | ctx->dops = &sockfs_dentry_operations; | |
364 | ctx->xattr = sockfs_xattr_handlers; | |
365 | return 0; | |
c74a1cbb AV |
366 | } |
367 | ||
368 | static struct vfsmount *sock_mnt __read_mostly; | |
369 | ||
370 | static struct file_system_type sock_fs_type = { | |
371 | .name = "sockfs", | |
fba9be49 | 372 | .init_fs_context = sockfs_init_fs_context, |
c74a1cbb AV |
373 | .kill_sb = kill_anon_super, |
374 | }; | |
375 | ||
1da177e4 LT |
376 | /* |
377 | * Obtains the first available file descriptor and sets it up for use. | |
378 | * | |
39d8c1b6 DM |
379 | * These functions create file structures and maps them to fd space |
380 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
381 | * and file struct implicitly stored in sock->file. |
382 | * Note that another thread may close file descriptor before we return | |
383 | * from this function. We use the fact that now we do not refer | |
384 | * to socket after mapping. If one day we will need it, this | |
385 | * function will increment ref. count on file by 1. | |
386 | * | |
387 | * In any case returned fd MAY BE not valid! | |
388 | * This race condition is unavoidable | |
389 | * with shared fd spaces, we cannot solve it inside kernel, | |
390 | * but we take care of internal coherence yet. | |
391 | */ | |
392 | ||
8a3c245c PT |
393 | /** |
394 | * sock_alloc_file - Bind a &socket to a &file | |
395 | * @sock: socket | |
396 | * @flags: file status flags | |
397 | * @dname: protocol name | |
398 | * | |
399 | * Returns the &file bound with @sock, implicitly storing it | |
400 | * in sock->file. If dname is %NULL, sets to "". | |
401 | * On failure the return is a ERR pointer (see linux/err.h). | |
402 | * This function uses GFP_KERNEL internally. | |
403 | */ | |
404 | ||
aab174f0 | 405 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 406 | { |
7cbe66b6 | 407 | struct file *file; |
1da177e4 | 408 | |
d93aa9d8 AV |
409 | if (!dname) |
410 | dname = sock->sk ? sock->sk->sk_prot_creator->name : ""; | |
39d8c1b6 | 411 | |
d93aa9d8 AV |
412 | file = alloc_file_pseudo(SOCK_INODE(sock), sock_mnt, dname, |
413 | O_RDWR | (flags & O_NONBLOCK), | |
414 | &socket_file_ops); | |
b5ffe634 | 415 | if (IS_ERR(file)) { |
8e1611e2 | 416 | sock_release(sock); |
39b65252 | 417 | return file; |
cc3808f8 AV |
418 | } |
419 | ||
420 | sock->file = file; | |
39d8c1b6 | 421 | file->private_data = sock; |
d8e464ec | 422 | stream_open(SOCK_INODE(sock), file); |
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); |
4546e44c | 437 | if (!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 | ||
8a3c245c PT |
446 | /** |
447 | * sock_from_file - Return the &socket bounded to @file. | |
448 | * @file: file | |
8a3c245c | 449 | * |
dba4a925 | 450 | * On failure returns %NULL. |
8a3c245c PT |
451 | */ |
452 | ||
dba4a925 | 453 | struct socket *sock_from_file(struct file *file) |
6cb153ca | 454 | { |
6cb153ca BL |
455 | if (file->f_op == &socket_file_ops) |
456 | return file->private_data; /* set in sock_map_fd */ | |
457 | ||
23bb80d2 | 458 | return NULL; |
6cb153ca | 459 | } |
406a3c63 | 460 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 461 | |
1da177e4 | 462 | /** |
c6d409cf | 463 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
464 | * @fd: file handle |
465 | * @err: pointer to an error code return | |
466 | * | |
467 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 468 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
469 | * with a negative errno code and NULL is returned. The function checks |
470 | * for both invalid handles and passing a handle which is not a socket. | |
471 | * | |
472 | * On a success the socket object pointer is returned. | |
473 | */ | |
474 | ||
475 | struct socket *sockfd_lookup(int fd, int *err) | |
476 | { | |
477 | struct file *file; | |
1da177e4 LT |
478 | struct socket *sock; |
479 | ||
89bddce5 SH |
480 | file = fget(fd); |
481 | if (!file) { | |
1da177e4 LT |
482 | *err = -EBADF; |
483 | return NULL; | |
484 | } | |
89bddce5 | 485 | |
dba4a925 FR |
486 | sock = sock_from_file(file); |
487 | if (!sock) { | |
488 | *err = -ENOTSOCK; | |
1da177e4 | 489 | fput(file); |
dba4a925 | 490 | } |
6cb153ca BL |
491 | return sock; |
492 | } | |
c6d409cf | 493 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 494 | |
6cb153ca BL |
495 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
496 | { | |
00e188ef | 497 | struct fd f = fdget(fd); |
6cb153ca BL |
498 | struct socket *sock; |
499 | ||
3672558c | 500 | *err = -EBADF; |
00e188ef | 501 | if (f.file) { |
dba4a925 | 502 | sock = sock_from_file(f.file); |
00e188ef | 503 | if (likely(sock)) { |
ce787a5a | 504 | *fput_needed = f.flags & FDPUT_FPUT; |
6cb153ca | 505 | return sock; |
00e188ef | 506 | } |
dba4a925 | 507 | *err = -ENOTSOCK; |
00e188ef | 508 | fdput(f); |
1da177e4 | 509 | } |
6cb153ca | 510 | return NULL; |
1da177e4 LT |
511 | } |
512 | ||
600e1779 MY |
513 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
514 | size_t size) | |
515 | { | |
516 | ssize_t len; | |
517 | ssize_t used = 0; | |
518 | ||
c5ef6035 | 519 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
520 | if (len < 0) |
521 | return len; | |
522 | used += len; | |
523 | if (buffer) { | |
524 | if (size < used) | |
525 | return -ERANGE; | |
526 | buffer += len; | |
527 | } | |
528 | ||
529 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
530 | used += len; | |
531 | if (buffer) { | |
532 | if (size < used) | |
533 | return -ERANGE; | |
534 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
535 | buffer += len; | |
536 | } | |
537 | ||
538 | return used; | |
539 | } | |
540 | ||
549c7297 CB |
541 | static int sockfs_setattr(struct user_namespace *mnt_userns, |
542 | struct dentry *dentry, struct iattr *iattr) | |
86741ec2 | 543 | { |
549c7297 | 544 | int err = simple_setattr(&init_user_ns, dentry, iattr); |
86741ec2 | 545 | |
e1a3a60a | 546 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
547 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
548 | ||
6d8c50dc CW |
549 | if (sock->sk) |
550 | sock->sk->sk_uid = iattr->ia_uid; | |
551 | else | |
552 | err = -ENOENT; | |
86741ec2 LC |
553 | } |
554 | ||
555 | return err; | |
556 | } | |
557 | ||
600e1779 | 558 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 559 | .listxattr = sockfs_listxattr, |
86741ec2 | 560 | .setattr = sockfs_setattr, |
600e1779 MY |
561 | }; |
562 | ||
1da177e4 | 563 | /** |
8a3c245c | 564 | * sock_alloc - allocate a socket |
89bddce5 | 565 | * |
1da177e4 LT |
566 | * Allocate a new inode and socket object. The two are bound together |
567 | * and initialised. The socket is then returned. If we are out of inodes | |
8a3c245c | 568 | * NULL is returned. This functions uses GFP_KERNEL internally. |
1da177e4 LT |
569 | */ |
570 | ||
f4a00aac | 571 | struct socket *sock_alloc(void) |
1da177e4 | 572 | { |
89bddce5 SH |
573 | struct inode *inode; |
574 | struct socket *sock; | |
1da177e4 | 575 | |
a209dfc7 | 576 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
577 | if (!inode) |
578 | return NULL; | |
579 | ||
580 | sock = SOCKET_I(inode); | |
581 | ||
85fe4025 | 582 | inode->i_ino = get_next_ino(); |
89bddce5 | 583 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
584 | inode->i_uid = current_fsuid(); |
585 | inode->i_gid = current_fsgid(); | |
600e1779 | 586 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 587 | |
1da177e4 LT |
588 | return sock; |
589 | } | |
f4a00aac | 590 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 591 | |
6d8c50dc | 592 | static void __sock_release(struct socket *sock, struct inode *inode) |
1da177e4 LT |
593 | { |
594 | if (sock->ops) { | |
595 | struct module *owner = sock->ops->owner; | |
596 | ||
6d8c50dc CW |
597 | if (inode) |
598 | inode_lock(inode); | |
1da177e4 | 599 | sock->ops->release(sock); |
ff7b11aa | 600 | sock->sk = NULL; |
6d8c50dc CW |
601 | if (inode) |
602 | inode_unlock(inode); | |
1da177e4 LT |
603 | sock->ops = NULL; |
604 | module_put(owner); | |
605 | } | |
606 | ||
333f7909 | 607 | if (sock->wq.fasync_list) |
3410f22e | 608 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 609 | |
1da177e4 LT |
610 | if (!sock->file) { |
611 | iput(SOCK_INODE(sock)); | |
612 | return; | |
613 | } | |
89bddce5 | 614 | sock->file = NULL; |
1da177e4 | 615 | } |
6d8c50dc | 616 | |
9a8ad9ac AL |
617 | /** |
618 | * sock_release - close a socket | |
619 | * @sock: socket to close | |
620 | * | |
621 | * The socket is released from the protocol stack if it has a release | |
622 | * callback, and the inode is then released if the socket is bound to | |
623 | * an inode not a file. | |
624 | */ | |
6d8c50dc CW |
625 | void sock_release(struct socket *sock) |
626 | { | |
627 | __sock_release(sock, NULL); | |
628 | } | |
c6d409cf | 629 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 630 | |
c14ac945 | 631 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 632 | { |
140c55d4 ED |
633 | u8 flags = *tx_flags; |
634 | ||
c14ac945 | 635 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
636 | flags |= SKBTX_HW_TSTAMP; |
637 | ||
c14ac945 | 638 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
639 | flags |= SKBTX_SW_TSTAMP; |
640 | ||
c14ac945 | 641 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
642 | flags |= SKBTX_SCHED_TSTAMP; |
643 | ||
140c55d4 | 644 | *tx_flags = flags; |
20d49473 | 645 | } |
67cc0d40 | 646 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 647 | |
8c3c447b PA |
648 | INDIRECT_CALLABLE_DECLARE(int inet_sendmsg(struct socket *, struct msghdr *, |
649 | size_t)); | |
a648a592 PA |
650 | INDIRECT_CALLABLE_DECLARE(int inet6_sendmsg(struct socket *, struct msghdr *, |
651 | size_t)); | |
d8725c86 | 652 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 653 | { |
a648a592 PA |
654 | int ret = INDIRECT_CALL_INET(sock->ops->sendmsg, inet6_sendmsg, |
655 | inet_sendmsg, sock, msg, | |
656 | msg_data_left(msg)); | |
d8725c86 AV |
657 | BUG_ON(ret == -EIOCBQUEUED); |
658 | return ret; | |
1da177e4 LT |
659 | } |
660 | ||
85806af0 RD |
661 | /** |
662 | * sock_sendmsg - send a message through @sock | |
663 | * @sock: socket | |
664 | * @msg: message to send | |
665 | * | |
666 | * Sends @msg through @sock, passing through LSM. | |
667 | * Returns the number of bytes sent, or an error code. | |
668 | */ | |
d8725c86 | 669 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 670 | { |
d8725c86 | 671 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 672 | msg_data_left(msg)); |
228e548e | 673 | |
d8725c86 | 674 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 675 | } |
c6d409cf | 676 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 | 677 | |
8a3c245c PT |
678 | /** |
679 | * kernel_sendmsg - send a message through @sock (kernel-space) | |
680 | * @sock: socket | |
681 | * @msg: message header | |
682 | * @vec: kernel vec | |
683 | * @num: vec array length | |
684 | * @size: total message data size | |
685 | * | |
686 | * Builds the message data with @vec and sends it through @sock. | |
687 | * Returns the number of bytes sent, or an error code. | |
688 | */ | |
689 | ||
1da177e4 LT |
690 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
691 | struct kvec *vec, size_t num, size_t size) | |
692 | { | |
aa563d7b | 693 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
d8725c86 | 694 | return sock_sendmsg(sock, msg); |
1da177e4 | 695 | } |
c6d409cf | 696 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 697 | |
8a3c245c PT |
698 | /** |
699 | * kernel_sendmsg_locked - send a message through @sock (kernel-space) | |
700 | * @sk: sock | |
701 | * @msg: message header | |
702 | * @vec: output s/g array | |
703 | * @num: output s/g array length | |
704 | * @size: total message data size | |
705 | * | |
706 | * Builds the message data with @vec and sends it through @sock. | |
707 | * Returns the number of bytes sent, or an error code. | |
708 | * Caller must hold @sk. | |
709 | */ | |
710 | ||
306b13eb TH |
711 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
712 | struct kvec *vec, size_t num, size_t size) | |
713 | { | |
714 | struct socket *sock = sk->sk_socket; | |
715 | ||
716 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 717 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb | 718 | |
aa563d7b | 719 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
306b13eb TH |
720 | |
721 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
722 | } | |
723 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
724 | ||
8605330a SHY |
725 | static bool skb_is_err_queue(const struct sk_buff *skb) |
726 | { | |
727 | /* pkt_type of skbs enqueued on the error queue are set to | |
728 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
729 | * in recvmsg, since skbs received on a local socket will never | |
730 | * have a pkt_type of PACKET_OUTGOING. | |
731 | */ | |
732 | return skb->pkt_type == PACKET_OUTGOING; | |
733 | } | |
734 | ||
b50a5c70 ML |
735 | /* On transmit, software and hardware timestamps are returned independently. |
736 | * As the two skb clones share the hardware timestamp, which may be updated | |
737 | * before the software timestamp is received, a hardware TX timestamp may be | |
738 | * returned only if there is no software TX timestamp. Ignore false software | |
739 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
7f1bc6e9 | 740 | * option SO_TIMESTAMP_OLD(NS) is enabled on the socket, even when the skb has a |
b50a5c70 ML |
741 | * hardware timestamp. |
742 | */ | |
743 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
744 | { | |
745 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
746 | } | |
747 | ||
aad9c8c4 ML |
748 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb) |
749 | { | |
750 | struct scm_ts_pktinfo ts_pktinfo; | |
751 | struct net_device *orig_dev; | |
752 | ||
753 | if (!skb_mac_header_was_set(skb)) | |
754 | return; | |
755 | ||
756 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
757 | ||
758 | rcu_read_lock(); | |
759 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
760 | if (orig_dev) | |
761 | ts_pktinfo.if_index = orig_dev->ifindex; | |
762 | rcu_read_unlock(); | |
763 | ||
764 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
765 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
766 | sizeof(ts_pktinfo), &ts_pktinfo); | |
767 | } | |
768 | ||
92f37fd2 ED |
769 | /* |
770 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
771 | */ | |
772 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
773 | struct sk_buff *skb) | |
774 | { | |
20d49473 | 775 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
887feae3 | 776 | int new_tstamp = sock_flag(sk, SOCK_TSTAMP_NEW); |
9718475e DD |
777 | struct scm_timestamping_internal tss; |
778 | ||
b50a5c70 | 779 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
780 | struct skb_shared_hwtstamps *shhwtstamps = |
781 | skb_hwtstamps(skb); | |
782 | ||
783 | /* Race occurred between timestamp enabling and packet | |
784 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 785 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 786 | __net_timestamp(skb); |
b50a5c70 ML |
787 | false_tstamp = 1; |
788 | } | |
20d49473 PO |
789 | |
790 | if (need_software_tstamp) { | |
791 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
887feae3 DD |
792 | if (new_tstamp) { |
793 | struct __kernel_sock_timeval tv; | |
794 | ||
795 | skb_get_new_timestamp(skb, &tv); | |
796 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_NEW, | |
797 | sizeof(tv), &tv); | |
798 | } else { | |
799 | struct __kernel_old_timeval tv; | |
800 | ||
801 | skb_get_timestamp(skb, &tv); | |
802 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_OLD, | |
803 | sizeof(tv), &tv); | |
804 | } | |
20d49473 | 805 | } else { |
887feae3 DD |
806 | if (new_tstamp) { |
807 | struct __kernel_timespec ts; | |
808 | ||
809 | skb_get_new_timestampns(skb, &ts); | |
810 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_NEW, | |
811 | sizeof(ts), &ts); | |
812 | } else { | |
df1b4ba9 | 813 | struct __kernel_old_timespec ts; |
887feae3 DD |
814 | |
815 | skb_get_timestampns(skb, &ts); | |
816 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_OLD, | |
817 | sizeof(ts), &ts); | |
818 | } | |
20d49473 PO |
819 | } |
820 | } | |
821 | ||
f24b9be5 | 822 | memset(&tss, 0, sizeof(tss)); |
c199105d | 823 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
9718475e | 824 | ktime_to_timespec64_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 825 | empty = 0; |
4d276eb6 | 826 | if (shhwtstamps && |
b9f40e21 | 827 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
b50a5c70 | 828 | !skb_is_swtx_tstamp(skb, false_tstamp) && |
9718475e | 829 | ktime_to_timespec64_cond(shhwtstamps->hwtstamp, tss.ts + 2)) { |
4d276eb6 | 830 | empty = 0; |
aad9c8c4 ML |
831 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && |
832 | !skb_is_err_queue(skb)) | |
833 | put_ts_pktinfo(msg, skb); | |
834 | } | |
1c885808 | 835 | if (!empty) { |
9718475e DD |
836 | if (sock_flag(sk, SOCK_TSTAMP_NEW)) |
837 | put_cmsg_scm_timestamping64(msg, &tss); | |
838 | else | |
839 | put_cmsg_scm_timestamping(msg, &tss); | |
1c885808 | 840 | |
8605330a | 841 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 842 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
843 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
844 | skb->len, skb->data); | |
845 | } | |
92f37fd2 | 846 | } |
7c81fd8b ACM |
847 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
848 | ||
6e3e939f JB |
849 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
850 | struct sk_buff *skb) | |
851 | { | |
852 | int ack; | |
853 | ||
854 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
855 | return; | |
856 | if (!skb->wifi_acked_valid) | |
857 | return; | |
858 | ||
859 | ack = skb->wifi_acked; | |
860 | ||
861 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
862 | } | |
863 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
864 | ||
11165f14 | 865 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
866 | struct sk_buff *skb) | |
3b885787 | 867 | { |
744d5a3e | 868 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 869 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 870 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
871 | } |
872 | ||
767dd033 | 873 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
874 | struct sk_buff *skb) |
875 | { | |
876 | sock_recv_timestamp(msg, sk, skb); | |
877 | sock_recv_drops(msg, sk, skb); | |
878 | } | |
767dd033 | 879 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 880 | |
8c3c447b | 881 | INDIRECT_CALLABLE_DECLARE(int inet_recvmsg(struct socket *, struct msghdr *, |
a648a592 PA |
882 | size_t, int)); |
883 | INDIRECT_CALLABLE_DECLARE(int inet6_recvmsg(struct socket *, struct msghdr *, | |
884 | size_t, int)); | |
1b784140 | 885 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 886 | int flags) |
1da177e4 | 887 | { |
a648a592 PA |
888 | return INDIRECT_CALL_INET(sock->ops->recvmsg, inet6_recvmsg, |
889 | inet_recvmsg, sock, msg, msg_data_left(msg), | |
890 | flags); | |
1da177e4 LT |
891 | } |
892 | ||
85806af0 RD |
893 | /** |
894 | * sock_recvmsg - receive a message from @sock | |
895 | * @sock: socket | |
896 | * @msg: message to receive | |
897 | * @flags: message flags | |
898 | * | |
899 | * Receives @msg from @sock, passing through LSM. Returns the total number | |
900 | * of bytes received, or an error. | |
901 | */ | |
2da62906 | 902 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 903 | { |
2da62906 | 904 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 905 | |
2da62906 | 906 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 907 | } |
c6d409cf | 908 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 909 | |
c1249c0a | 910 | /** |
8a3c245c PT |
911 | * kernel_recvmsg - Receive a message from a socket (kernel space) |
912 | * @sock: The socket to receive the message from | |
913 | * @msg: Received message | |
914 | * @vec: Input s/g array for message data | |
915 | * @num: Size of input s/g array | |
916 | * @size: Number of bytes to read | |
917 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
c1249c0a | 918 | * |
8a3c245c PT |
919 | * On return the msg structure contains the scatter/gather array passed in the |
920 | * vec argument. The array is modified so that it consists of the unfilled | |
921 | * portion of the original array. | |
c1249c0a | 922 | * |
8a3c245c | 923 | * The returned value is the total number of bytes received, or an error. |
c1249c0a | 924 | */ |
8a3c245c | 925 | |
89bddce5 SH |
926 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
927 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 | 928 | { |
1f466e1f | 929 | msg->msg_control_is_user = false; |
aa563d7b | 930 | iov_iter_kvec(&msg->msg_iter, READ, vec, num, size); |
1f466e1f | 931 | return sock_recvmsg(sock, msg, flags); |
1da177e4 | 932 | } |
c6d409cf | 933 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 934 | |
ce1d4d3e CH |
935 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
936 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 937 | { |
1da177e4 LT |
938 | struct socket *sock; |
939 | int flags; | |
940 | ||
ce1d4d3e CH |
941 | sock = file->private_data; |
942 | ||
35f9c09f ED |
943 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
944 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
945 | flags |= more; | |
ce1d4d3e | 946 | |
e6949583 | 947 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 948 | } |
1da177e4 | 949 | |
9c55e01c | 950 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 951 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
952 | unsigned int flags) |
953 | { | |
954 | struct socket *sock = file->private_data; | |
955 | ||
997b37da | 956 | if (unlikely(!sock->ops->splice_read)) |
95506588 | 957 | return generic_file_splice_read(file, ppos, pipe, len, flags); |
997b37da | 958 | |
9c55e01c JA |
959 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
960 | } | |
961 | ||
8ae5e030 | 962 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 963 | { |
6d652330 AV |
964 | struct file *file = iocb->ki_filp; |
965 | struct socket *sock = file->private_data; | |
0345f931 | 966 | struct msghdr msg = {.msg_iter = *to, |
967 | .msg_iocb = iocb}; | |
8ae5e030 | 968 | ssize_t res; |
ce1d4d3e | 969 | |
ebfcd895 | 970 | if (file->f_flags & O_NONBLOCK || (iocb->ki_flags & IOCB_NOWAIT)) |
8ae5e030 AV |
971 | msg.msg_flags = MSG_DONTWAIT; |
972 | ||
973 | if (iocb->ki_pos != 0) | |
1da177e4 | 974 | return -ESPIPE; |
027445c3 | 975 | |
66ee59af | 976 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
977 | return 0; |
978 | ||
2da62906 | 979 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
980 | *to = msg.msg_iter; |
981 | return res; | |
1da177e4 LT |
982 | } |
983 | ||
8ae5e030 | 984 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 985 | { |
6d652330 AV |
986 | struct file *file = iocb->ki_filp; |
987 | struct socket *sock = file->private_data; | |
0345f931 | 988 | struct msghdr msg = {.msg_iter = *from, |
989 | .msg_iocb = iocb}; | |
8ae5e030 | 990 | ssize_t res; |
1da177e4 | 991 | |
8ae5e030 | 992 | if (iocb->ki_pos != 0) |
ce1d4d3e | 993 | return -ESPIPE; |
027445c3 | 994 | |
ebfcd895 | 995 | if (file->f_flags & O_NONBLOCK || (iocb->ki_flags & IOCB_NOWAIT)) |
8ae5e030 AV |
996 | msg.msg_flags = MSG_DONTWAIT; |
997 | ||
6d652330 AV |
998 | if (sock->type == SOCK_SEQPACKET) |
999 | msg.msg_flags |= MSG_EOR; | |
1000 | ||
d8725c86 | 1001 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
1002 | *from = msg.msg_iter; |
1003 | return res; | |
1da177e4 LT |
1004 | } |
1005 | ||
1da177e4 LT |
1006 | /* |
1007 | * Atomic setting of ioctl hooks to avoid race | |
1008 | * with module unload. | |
1009 | */ | |
1010 | ||
4a3e2f71 | 1011 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 1012 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 1013 | |
881d966b | 1014 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 1015 | { |
4a3e2f71 | 1016 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 1017 | br_ioctl_hook = hook; |
4a3e2f71 | 1018 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1019 | } |
1020 | EXPORT_SYMBOL(brioctl_set); | |
1021 | ||
4a3e2f71 | 1022 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 1023 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 1024 | |
881d966b | 1025 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 1026 | { |
4a3e2f71 | 1027 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1028 | vlan_ioctl_hook = hook; |
4a3e2f71 | 1029 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
1030 | } |
1031 | EXPORT_SYMBOL(vlan_ioctl_set); | |
1032 | ||
6b96018b | 1033 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
63ff03ab | 1034 | unsigned int cmd, unsigned long arg) |
6b96018b AB |
1035 | { |
1036 | int err; | |
1037 | void __user *argp = (void __user *)arg; | |
1038 | ||
1039 | err = sock->ops->ioctl(sock, cmd, arg); | |
1040 | ||
1041 | /* | |
1042 | * If this ioctl is unknown try to hand it down | |
1043 | * to the NIC driver. | |
1044 | */ | |
36fd633e AV |
1045 | if (err != -ENOIOCTLCMD) |
1046 | return err; | |
6b96018b | 1047 | |
36fd633e AV |
1048 | if (cmd == SIOCGIFCONF) { |
1049 | struct ifconf ifc; | |
1050 | if (copy_from_user(&ifc, argp, sizeof(struct ifconf))) | |
1051 | return -EFAULT; | |
1052 | rtnl_lock(); | |
1053 | err = dev_ifconf(net, &ifc, sizeof(struct ifreq)); | |
1054 | rtnl_unlock(); | |
1055 | if (!err && copy_to_user(argp, &ifc, sizeof(struct ifconf))) | |
1056 | err = -EFAULT; | |
44c02a2c AV |
1057 | } else { |
1058 | struct ifreq ifr; | |
1059 | bool need_copyout; | |
63ff03ab | 1060 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) |
44c02a2c AV |
1061 | return -EFAULT; |
1062 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1063 | if (!err && need_copyout) | |
63ff03ab | 1064 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) |
44c02a2c | 1065 | return -EFAULT; |
36fd633e | 1066 | } |
6b96018b AB |
1067 | return err; |
1068 | } | |
1069 | ||
1da177e4 LT |
1070 | /* |
1071 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1072 | * what to do with it - that's up to the protocol still. | |
1073 | */ | |
1074 | ||
8a3c245c PT |
1075 | /** |
1076 | * get_net_ns - increment the refcount of the network namespace | |
1077 | * @ns: common namespace (net) | |
1078 | * | |
1079 | * Returns the net's common namespace. | |
1080 | */ | |
1081 | ||
d8d211a2 | 1082 | struct ns_common *get_net_ns(struct ns_common *ns) |
c62cce2c AV |
1083 | { |
1084 | return &get_net(container_of(ns, struct net, ns))->ns; | |
1085 | } | |
d8d211a2 | 1086 | EXPORT_SYMBOL_GPL(get_net_ns); |
c62cce2c | 1087 | |
1da177e4 LT |
1088 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
1089 | { | |
1090 | struct socket *sock; | |
881d966b | 1091 | struct sock *sk; |
1da177e4 LT |
1092 | void __user *argp = (void __user *)arg; |
1093 | int pid, err; | |
881d966b | 1094 | struct net *net; |
1da177e4 | 1095 | |
b69aee04 | 1096 | sock = file->private_data; |
881d966b | 1097 | sk = sock->sk; |
3b1e0a65 | 1098 | net = sock_net(sk); |
44c02a2c AV |
1099 | if (unlikely(cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))) { |
1100 | struct ifreq ifr; | |
1101 | bool need_copyout; | |
1102 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
1103 | return -EFAULT; | |
1104 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1105 | if (!err && need_copyout) | |
1106 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
1107 | return -EFAULT; | |
1da177e4 | 1108 | } else |
3d23e349 | 1109 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1110 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
b1b0c245 | 1111 | err = wext_handle_ioctl(net, cmd, argp); |
1da177e4 | 1112 | } else |
3d23e349 | 1113 | #endif |
89bddce5 | 1114 | switch (cmd) { |
1da177e4 LT |
1115 | case FIOSETOWN: |
1116 | case SIOCSPGRP: | |
1117 | err = -EFAULT; | |
1118 | if (get_user(pid, (int __user *)argp)) | |
1119 | break; | |
393cc3f5 | 1120 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1121 | break; |
1122 | case FIOGETOWN: | |
1123 | case SIOCGPGRP: | |
609d7fa9 | 1124 | err = put_user(f_getown(sock->file), |
89bddce5 | 1125 | (int __user *)argp); |
1da177e4 LT |
1126 | break; |
1127 | case SIOCGIFBR: | |
1128 | case SIOCSIFBR: | |
1129 | case SIOCBRADDBR: | |
1130 | case SIOCBRDELBR: | |
1131 | err = -ENOPKG; | |
1132 | if (!br_ioctl_hook) | |
1133 | request_module("bridge"); | |
1134 | ||
4a3e2f71 | 1135 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1136 | if (br_ioctl_hook) |
881d966b | 1137 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1138 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1139 | break; |
1140 | case SIOCGIFVLAN: | |
1141 | case SIOCSIFVLAN: | |
1142 | err = -ENOPKG; | |
1143 | if (!vlan_ioctl_hook) | |
1144 | request_module("8021q"); | |
1145 | ||
4a3e2f71 | 1146 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1147 | if (vlan_ioctl_hook) |
881d966b | 1148 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1149 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1150 | break; |
c62cce2c AV |
1151 | case SIOCGSKNS: |
1152 | err = -EPERM; | |
1153 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1154 | break; | |
1155 | ||
1156 | err = open_related_ns(&net->ns, get_net_ns); | |
1157 | break; | |
0768e170 AB |
1158 | case SIOCGSTAMP_OLD: |
1159 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
1160 | if (!sock->ops->gettstamp) { |
1161 | err = -ENOIOCTLCMD; | |
1162 | break; | |
1163 | } | |
1164 | err = sock->ops->gettstamp(sock, argp, | |
0768e170 AB |
1165 | cmd == SIOCGSTAMP_OLD, |
1166 | !IS_ENABLED(CONFIG_64BIT)); | |
60747828 | 1167 | break; |
0768e170 AB |
1168 | case SIOCGSTAMP_NEW: |
1169 | case SIOCGSTAMPNS_NEW: | |
1170 | if (!sock->ops->gettstamp) { | |
1171 | err = -ENOIOCTLCMD; | |
1172 | break; | |
1173 | } | |
1174 | err = sock->ops->gettstamp(sock, argp, | |
1175 | cmd == SIOCGSTAMP_NEW, | |
1176 | false); | |
c7cbdbf2 | 1177 | break; |
1da177e4 | 1178 | default: |
63ff03ab | 1179 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1180 | break; |
89bddce5 | 1181 | } |
1da177e4 LT |
1182 | return err; |
1183 | } | |
1184 | ||
8a3c245c PT |
1185 | /** |
1186 | * sock_create_lite - creates a socket | |
1187 | * @family: protocol family (AF_INET, ...) | |
1188 | * @type: communication type (SOCK_STREAM, ...) | |
1189 | * @protocol: protocol (0, ...) | |
1190 | * @res: new socket | |
1191 | * | |
1192 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1193 | * The new socket initialization is not complete, see kernel_accept(). | |
1194 | * Returns 0 or an error. On failure @res is set to %NULL. | |
1195 | * This function internally uses GFP_KERNEL. | |
1196 | */ | |
1197 | ||
1da177e4 LT |
1198 | int sock_create_lite(int family, int type, int protocol, struct socket **res) |
1199 | { | |
1200 | int err; | |
1201 | struct socket *sock = NULL; | |
89bddce5 | 1202 | |
1da177e4 LT |
1203 | err = security_socket_create(family, type, protocol, 1); |
1204 | if (err) | |
1205 | goto out; | |
1206 | ||
1207 | sock = sock_alloc(); | |
1208 | if (!sock) { | |
1209 | err = -ENOMEM; | |
1210 | goto out; | |
1211 | } | |
1212 | ||
1da177e4 | 1213 | sock->type = type; |
7420ed23 VY |
1214 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1215 | if (err) | |
1216 | goto out_release; | |
1217 | ||
1da177e4 LT |
1218 | out: |
1219 | *res = sock; | |
1220 | return err; | |
7420ed23 VY |
1221 | out_release: |
1222 | sock_release(sock); | |
1223 | sock = NULL; | |
1224 | goto out; | |
1da177e4 | 1225 | } |
c6d409cf | 1226 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1227 | |
1228 | /* No kernel lock held - perfect */ | |
ade994f4 | 1229 | static __poll_t sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1230 | { |
3cafb376 | 1231 | struct socket *sock = file->private_data; |
a331de3b | 1232 | __poll_t events = poll_requested_events(wait), flag = 0; |
2d48d67f | 1233 | |
e88958e6 CH |
1234 | if (!sock->ops->poll) |
1235 | return 0; | |
f641f13b | 1236 | |
a331de3b CH |
1237 | if (sk_can_busy_loop(sock->sk)) { |
1238 | /* poll once if requested by the syscall */ | |
1239 | if (events & POLL_BUSY_LOOP) | |
1240 | sk_busy_loop(sock->sk, 1); | |
1241 | ||
1242 | /* if this socket can poll_ll, tell the system call */ | |
1243 | flag = POLL_BUSY_LOOP; | |
1244 | } | |
1245 | ||
1246 | return sock->ops->poll(file, sock, wait) | flag; | |
1da177e4 LT |
1247 | } |
1248 | ||
89bddce5 | 1249 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1250 | { |
b69aee04 | 1251 | struct socket *sock = file->private_data; |
1da177e4 LT |
1252 | |
1253 | return sock->ops->mmap(file, sock, vma); | |
1254 | } | |
1255 | ||
20380731 | 1256 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1257 | { |
6d8c50dc | 1258 | __sock_release(SOCKET_I(inode), inode); |
1da177e4 LT |
1259 | return 0; |
1260 | } | |
1261 | ||
1262 | /* | |
1263 | * Update the socket async list | |
1264 | * | |
1265 | * Fasync_list locking strategy. | |
1266 | * | |
1267 | * 1. fasync_list is modified only under process context socket lock | |
1268 | * i.e. under semaphore. | |
1269 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1270 | * or under socket lock |
1da177e4 LT |
1271 | */ |
1272 | ||
1273 | static int sock_fasync(int fd, struct file *filp, int on) | |
1274 | { | |
989a2979 ED |
1275 | struct socket *sock = filp->private_data; |
1276 | struct sock *sk = sock->sk; | |
333f7909 | 1277 | struct socket_wq *wq = &sock->wq; |
1da177e4 | 1278 | |
989a2979 | 1279 | if (sk == NULL) |
1da177e4 | 1280 | return -EINVAL; |
1da177e4 LT |
1281 | |
1282 | lock_sock(sk); | |
eaefd110 | 1283 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1284 | |
eaefd110 | 1285 | if (!wq->fasync_list) |
989a2979 ED |
1286 | sock_reset_flag(sk, SOCK_FASYNC); |
1287 | else | |
bcdce719 | 1288 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1289 | |
989a2979 | 1290 | release_sock(sk); |
1da177e4 LT |
1291 | return 0; |
1292 | } | |
1293 | ||
ceb5d58b | 1294 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1295 | |
ceb5d58b | 1296 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1297 | { |
ceb5d58b | 1298 | if (!wq || !wq->fasync_list) |
1da177e4 | 1299 | return -1; |
ceb5d58b | 1300 | |
89bddce5 | 1301 | switch (how) { |
8d8ad9d7 | 1302 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1303 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1304 | break; |
1305 | goto call_kill; | |
8d8ad9d7 | 1306 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1307 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 | 1308 | break; |
7c7ab580 | 1309 | fallthrough; |
8d8ad9d7 | 1310 | case SOCK_WAKE_IO: |
89bddce5 | 1311 | call_kill: |
43815482 | 1312 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1313 | break; |
8d8ad9d7 | 1314 | case SOCK_WAKE_URG: |
43815482 | 1315 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1316 | } |
ceb5d58b | 1317 | |
1da177e4 LT |
1318 | return 0; |
1319 | } | |
c6d409cf | 1320 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1321 | |
8a3c245c PT |
1322 | /** |
1323 | * __sock_create - creates a socket | |
1324 | * @net: net namespace | |
1325 | * @family: protocol family (AF_INET, ...) | |
1326 | * @type: communication type (SOCK_STREAM, ...) | |
1327 | * @protocol: protocol (0, ...) | |
1328 | * @res: new socket | |
1329 | * @kern: boolean for kernel space sockets | |
1330 | * | |
1331 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1332 | * Returns 0 or an error. On failure @res is set to %NULL. @kern must | |
1333 | * be set to true if the socket resides in kernel space. | |
1334 | * This function internally uses GFP_KERNEL. | |
1335 | */ | |
1336 | ||
721db93a | 1337 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1338 | struct socket **res, int kern) |
1da177e4 LT |
1339 | { |
1340 | int err; | |
1341 | struct socket *sock; | |
55737fda | 1342 | const struct net_proto_family *pf; |
1da177e4 LT |
1343 | |
1344 | /* | |
89bddce5 | 1345 | * Check protocol is in range |
1da177e4 LT |
1346 | */ |
1347 | if (family < 0 || family >= NPROTO) | |
1348 | return -EAFNOSUPPORT; | |
1349 | if (type < 0 || type >= SOCK_MAX) | |
1350 | return -EINVAL; | |
1351 | ||
1352 | /* Compatibility. | |
1353 | ||
1354 | This uglymoron is moved from INET layer to here to avoid | |
1355 | deadlock in module load. | |
1356 | */ | |
1357 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1358 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1359 | current->comm); | |
1da177e4 LT |
1360 | family = PF_PACKET; |
1361 | } | |
1362 | ||
1363 | err = security_socket_create(family, type, protocol, kern); | |
1364 | if (err) | |
1365 | return err; | |
89bddce5 | 1366 | |
55737fda SH |
1367 | /* |
1368 | * Allocate the socket and allow the family to set things up. if | |
1369 | * the protocol is 0, the family is instructed to select an appropriate | |
1370 | * default. | |
1371 | */ | |
1372 | sock = sock_alloc(); | |
1373 | if (!sock) { | |
e87cc472 | 1374 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1375 | return -ENFILE; /* Not exactly a match, but its the |
1376 | closest posix thing */ | |
1377 | } | |
1378 | ||
1379 | sock->type = type; | |
1380 | ||
95a5afca | 1381 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1382 | /* Attempt to load a protocol module if the find failed. |
1383 | * | |
1384 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1385 | * requested real, full-featured networking support upon configuration. |
1386 | * Otherwise module support will break! | |
1387 | */ | |
190683a9 | 1388 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1389 | request_module("net-pf-%d", family); |
1da177e4 LT |
1390 | #endif |
1391 | ||
55737fda SH |
1392 | rcu_read_lock(); |
1393 | pf = rcu_dereference(net_families[family]); | |
1394 | err = -EAFNOSUPPORT; | |
1395 | if (!pf) | |
1396 | goto out_release; | |
1da177e4 LT |
1397 | |
1398 | /* | |
1399 | * We will call the ->create function, that possibly is in a loadable | |
1400 | * module, so we have to bump that loadable module refcnt first. | |
1401 | */ | |
55737fda | 1402 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1403 | goto out_release; |
1404 | ||
55737fda SH |
1405 | /* Now protected by module ref count */ |
1406 | rcu_read_unlock(); | |
1407 | ||
3f378b68 | 1408 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1409 | if (err < 0) |
1da177e4 | 1410 | goto out_module_put; |
a79af59e | 1411 | |
1da177e4 LT |
1412 | /* |
1413 | * Now to bump the refcnt of the [loadable] module that owns this | |
1414 | * socket at sock_release time we decrement its refcnt. | |
1415 | */ | |
55737fda SH |
1416 | if (!try_module_get(sock->ops->owner)) |
1417 | goto out_module_busy; | |
1418 | ||
1da177e4 LT |
1419 | /* |
1420 | * Now that we're done with the ->create function, the [loadable] | |
1421 | * module can have its refcnt decremented | |
1422 | */ | |
55737fda | 1423 | module_put(pf->owner); |
7420ed23 VY |
1424 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1425 | if (err) | |
3b185525 | 1426 | goto out_sock_release; |
55737fda | 1427 | *res = sock; |
1da177e4 | 1428 | |
55737fda SH |
1429 | return 0; |
1430 | ||
1431 | out_module_busy: | |
1432 | err = -EAFNOSUPPORT; | |
1da177e4 | 1433 | out_module_put: |
55737fda SH |
1434 | sock->ops = NULL; |
1435 | module_put(pf->owner); | |
1436 | out_sock_release: | |
1da177e4 | 1437 | sock_release(sock); |
55737fda SH |
1438 | return err; |
1439 | ||
1440 | out_release: | |
1441 | rcu_read_unlock(); | |
1442 | goto out_sock_release; | |
1da177e4 | 1443 | } |
721db93a | 1444 | EXPORT_SYMBOL(__sock_create); |
1da177e4 | 1445 | |
8a3c245c PT |
1446 | /** |
1447 | * sock_create - creates a socket | |
1448 | * @family: protocol family (AF_INET, ...) | |
1449 | * @type: communication type (SOCK_STREAM, ...) | |
1450 | * @protocol: protocol (0, ...) | |
1451 | * @res: new socket | |
1452 | * | |
1453 | * A wrapper around __sock_create(). | |
1454 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1455 | */ | |
1456 | ||
1da177e4 LT |
1457 | int sock_create(int family, int type, int protocol, struct socket **res) |
1458 | { | |
1b8d7ae4 | 1459 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1460 | } |
c6d409cf | 1461 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1462 | |
8a3c245c PT |
1463 | /** |
1464 | * sock_create_kern - creates a socket (kernel space) | |
1465 | * @net: net namespace | |
1466 | * @family: protocol family (AF_INET, ...) | |
1467 | * @type: communication type (SOCK_STREAM, ...) | |
1468 | * @protocol: protocol (0, ...) | |
1469 | * @res: new socket | |
1470 | * | |
1471 | * A wrapper around __sock_create(). | |
1472 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1473 | */ | |
1474 | ||
eeb1bd5c | 1475 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1476 | { |
eeb1bd5c | 1477 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1478 | } |
c6d409cf | 1479 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1480 | |
9d6a15c3 | 1481 | int __sys_socket(int family, int type, int protocol) |
1da177e4 LT |
1482 | { |
1483 | int retval; | |
1484 | struct socket *sock; | |
a677a039 UD |
1485 | int flags; |
1486 | ||
e38b36f3 UD |
1487 | /* Check the SOCK_* constants for consistency. */ |
1488 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1489 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1490 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1491 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1492 | ||
a677a039 | 1493 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1494 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1495 | return -EINVAL; |
1496 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1497 | |
aaca0bdc UD |
1498 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1499 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1500 | ||
1da177e4 LT |
1501 | retval = sock_create(family, type, protocol, &sock); |
1502 | if (retval < 0) | |
8e1611e2 | 1503 | return retval; |
1da177e4 | 1504 | |
8e1611e2 | 1505 | return sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1506 | } |
1507 | ||
9d6a15c3 DB |
1508 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1509 | { | |
1510 | return __sys_socket(family, type, protocol); | |
1511 | } | |
1512 | ||
1da177e4 LT |
1513 | /* |
1514 | * Create a pair of connected sockets. | |
1515 | */ | |
1516 | ||
6debc8d8 | 1517 | int __sys_socketpair(int family, int type, int protocol, int __user *usockvec) |
1da177e4 LT |
1518 | { |
1519 | struct socket *sock1, *sock2; | |
1520 | int fd1, fd2, err; | |
db349509 | 1521 | struct file *newfile1, *newfile2; |
a677a039 UD |
1522 | int flags; |
1523 | ||
1524 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1525 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1526 | return -EINVAL; |
1527 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1528 | |
aaca0bdc UD |
1529 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1530 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1531 | ||
016a266b AV |
1532 | /* |
1533 | * reserve descriptors and make sure we won't fail | |
1534 | * to return them to userland. | |
1535 | */ | |
1536 | fd1 = get_unused_fd_flags(flags); | |
1537 | if (unlikely(fd1 < 0)) | |
1538 | return fd1; | |
1539 | ||
1540 | fd2 = get_unused_fd_flags(flags); | |
1541 | if (unlikely(fd2 < 0)) { | |
1542 | put_unused_fd(fd1); | |
1543 | return fd2; | |
1544 | } | |
1545 | ||
1546 | err = put_user(fd1, &usockvec[0]); | |
1547 | if (err) | |
1548 | goto out; | |
1549 | ||
1550 | err = put_user(fd2, &usockvec[1]); | |
1551 | if (err) | |
1552 | goto out; | |
1553 | ||
1da177e4 LT |
1554 | /* |
1555 | * Obtain the first socket and check if the underlying protocol | |
1556 | * supports the socketpair call. | |
1557 | */ | |
1558 | ||
1559 | err = sock_create(family, type, protocol, &sock1); | |
016a266b | 1560 | if (unlikely(err < 0)) |
1da177e4 LT |
1561 | goto out; |
1562 | ||
1563 | err = sock_create(family, type, protocol, &sock2); | |
016a266b AV |
1564 | if (unlikely(err < 0)) { |
1565 | sock_release(sock1); | |
1566 | goto out; | |
bf3c23d1 | 1567 | } |
d73aa286 | 1568 | |
d47cd945 DH |
1569 | err = security_socket_socketpair(sock1, sock2); |
1570 | if (unlikely(err)) { | |
1571 | sock_release(sock2); | |
1572 | sock_release(sock1); | |
1573 | goto out; | |
1574 | } | |
1575 | ||
016a266b AV |
1576 | err = sock1->ops->socketpair(sock1, sock2); |
1577 | if (unlikely(err < 0)) { | |
1578 | sock_release(sock2); | |
1579 | sock_release(sock1); | |
1580 | goto out; | |
28407630 AV |
1581 | } |
1582 | ||
aab174f0 | 1583 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1584 | if (IS_ERR(newfile1)) { |
28407630 | 1585 | err = PTR_ERR(newfile1); |
016a266b AV |
1586 | sock_release(sock2); |
1587 | goto out; | |
28407630 AV |
1588 | } |
1589 | ||
aab174f0 | 1590 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1591 | if (IS_ERR(newfile2)) { |
1592 | err = PTR_ERR(newfile2); | |
016a266b AV |
1593 | fput(newfile1); |
1594 | goto out; | |
db349509 AV |
1595 | } |
1596 | ||
157cf649 | 1597 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1598 | |
db349509 AV |
1599 | fd_install(fd1, newfile1); |
1600 | fd_install(fd2, newfile2); | |
d73aa286 | 1601 | return 0; |
1da177e4 | 1602 | |
016a266b | 1603 | out: |
d73aa286 | 1604 | put_unused_fd(fd2); |
d73aa286 | 1605 | put_unused_fd(fd1); |
1da177e4 LT |
1606 | return err; |
1607 | } | |
1608 | ||
6debc8d8 DB |
1609 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1610 | int __user *, usockvec) | |
1611 | { | |
1612 | return __sys_socketpair(family, type, protocol, usockvec); | |
1613 | } | |
1614 | ||
1da177e4 LT |
1615 | /* |
1616 | * Bind a name to a socket. Nothing much to do here since it's | |
1617 | * the protocol's responsibility to handle the local address. | |
1618 | * | |
1619 | * We move the socket address to kernel space before we call | |
1620 | * the protocol layer (having also checked the address is ok). | |
1621 | */ | |
1622 | ||
a87d35d8 | 1623 | int __sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
1da177e4 LT |
1624 | { |
1625 | struct socket *sock; | |
230b1839 | 1626 | struct sockaddr_storage address; |
6cb153ca | 1627 | int err, fput_needed; |
1da177e4 | 1628 | |
89bddce5 | 1629 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1630 | if (sock) { |
43db362d | 1631 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
068b88cc | 1632 | if (!err) { |
89bddce5 | 1633 | err = security_socket_bind(sock, |
230b1839 | 1634 | (struct sockaddr *)&address, |
89bddce5 | 1635 | addrlen); |
6cb153ca BL |
1636 | if (!err) |
1637 | err = sock->ops->bind(sock, | |
89bddce5 | 1638 | (struct sockaddr *) |
230b1839 | 1639 | &address, addrlen); |
1da177e4 | 1640 | } |
6cb153ca | 1641 | fput_light(sock->file, fput_needed); |
89bddce5 | 1642 | } |
1da177e4 LT |
1643 | return err; |
1644 | } | |
1645 | ||
a87d35d8 DB |
1646 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1647 | { | |
1648 | return __sys_bind(fd, umyaddr, addrlen); | |
1649 | } | |
1650 | ||
1da177e4 LT |
1651 | /* |
1652 | * Perform a listen. Basically, we allow the protocol to do anything | |
1653 | * necessary for a listen, and if that works, we mark the socket as | |
1654 | * ready for listening. | |
1655 | */ | |
1656 | ||
25e290ee | 1657 | int __sys_listen(int fd, int backlog) |
1da177e4 LT |
1658 | { |
1659 | struct socket *sock; | |
6cb153ca | 1660 | int err, fput_needed; |
b8e1f9b5 | 1661 | int somaxconn; |
89bddce5 SH |
1662 | |
1663 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1664 | if (sock) { | |
8efa6e93 | 1665 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1666 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1667 | backlog = somaxconn; |
1da177e4 LT |
1668 | |
1669 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1670 | if (!err) |
1671 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1672 | |
6cb153ca | 1673 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1674 | } |
1675 | return err; | |
1676 | } | |
1677 | ||
25e290ee DB |
1678 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1679 | { | |
1680 | return __sys_listen(fd, backlog); | |
1681 | } | |
1682 | ||
de2ea4b6 JA |
1683 | int __sys_accept4_file(struct file *file, unsigned file_flags, |
1684 | struct sockaddr __user *upeer_sockaddr, | |
09952e3e JA |
1685 | int __user *upeer_addrlen, int flags, |
1686 | unsigned long nofile) | |
1da177e4 LT |
1687 | { |
1688 | struct socket *sock, *newsock; | |
39d8c1b6 | 1689 | struct file *newfile; |
de2ea4b6 | 1690 | int err, len, newfd; |
230b1839 | 1691 | struct sockaddr_storage address; |
1da177e4 | 1692 | |
77d27200 | 1693 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1694 | return -EINVAL; |
1695 | ||
1696 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1697 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1698 | ||
dba4a925 FR |
1699 | sock = sock_from_file(file); |
1700 | if (!sock) { | |
1701 | err = -ENOTSOCK; | |
1da177e4 | 1702 | goto out; |
dba4a925 | 1703 | } |
1da177e4 LT |
1704 | |
1705 | err = -ENFILE; | |
c6d409cf ED |
1706 | newsock = sock_alloc(); |
1707 | if (!newsock) | |
de2ea4b6 | 1708 | goto out; |
1da177e4 LT |
1709 | |
1710 | newsock->type = sock->type; | |
1711 | newsock->ops = sock->ops; | |
1712 | ||
1da177e4 LT |
1713 | /* |
1714 | * We don't need try_module_get here, as the listening socket (sock) | |
1715 | * has the protocol module (sock->ops->owner) held. | |
1716 | */ | |
1717 | __module_get(newsock->ops->owner); | |
1718 | ||
09952e3e | 1719 | newfd = __get_unused_fd_flags(flags, nofile); |
39d8c1b6 DM |
1720 | if (unlikely(newfd < 0)) { |
1721 | err = newfd; | |
9a1875e6 | 1722 | sock_release(newsock); |
de2ea4b6 | 1723 | goto out; |
39d8c1b6 | 1724 | } |
aab174f0 | 1725 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1726 | if (IS_ERR(newfile)) { |
28407630 AV |
1727 | err = PTR_ERR(newfile); |
1728 | put_unused_fd(newfd); | |
de2ea4b6 | 1729 | goto out; |
28407630 | 1730 | } |
39d8c1b6 | 1731 | |
a79af59e FF |
1732 | err = security_socket_accept(sock, newsock); |
1733 | if (err) | |
39d8c1b6 | 1734 | goto out_fd; |
a79af59e | 1735 | |
de2ea4b6 JA |
1736 | err = sock->ops->accept(sock, newsock, sock->file->f_flags | file_flags, |
1737 | false); | |
1da177e4 | 1738 | if (err < 0) |
39d8c1b6 | 1739 | goto out_fd; |
1da177e4 LT |
1740 | |
1741 | if (upeer_sockaddr) { | |
9b2c45d4 DV |
1742 | len = newsock->ops->getname(newsock, |
1743 | (struct sockaddr *)&address, 2); | |
1744 | if (len < 0) { | |
1da177e4 | 1745 | err = -ECONNABORTED; |
39d8c1b6 | 1746 | goto out_fd; |
1da177e4 | 1747 | } |
43db362d | 1748 | err = move_addr_to_user(&address, |
230b1839 | 1749 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1750 | if (err < 0) |
39d8c1b6 | 1751 | goto out_fd; |
1da177e4 LT |
1752 | } |
1753 | ||
1754 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1755 | ||
39d8c1b6 DM |
1756 | fd_install(newfd, newfile); |
1757 | err = newfd; | |
1da177e4 LT |
1758 | out: |
1759 | return err; | |
39d8c1b6 | 1760 | out_fd: |
9606a216 | 1761 | fput(newfile); |
39d8c1b6 | 1762 | put_unused_fd(newfd); |
de2ea4b6 JA |
1763 | goto out; |
1764 | ||
1765 | } | |
1766 | ||
1767 | /* | |
1768 | * For accept, we attempt to create a new socket, set up the link | |
1769 | * with the client, wake up the client, then return the new | |
1770 | * connected fd. We collect the address of the connector in kernel | |
1771 | * space and move it to user at the very end. This is unclean because | |
1772 | * we open the socket then return an error. | |
1773 | * | |
1774 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1775 | * status to recvmsg. We need to add that support in a way thats | |
1776 | * clean when we restructure accept also. | |
1777 | */ | |
1778 | ||
1779 | int __sys_accept4(int fd, struct sockaddr __user *upeer_sockaddr, | |
1780 | int __user *upeer_addrlen, int flags) | |
1781 | { | |
1782 | int ret = -EBADF; | |
1783 | struct fd f; | |
1784 | ||
1785 | f = fdget(fd); | |
1786 | if (f.file) { | |
1787 | ret = __sys_accept4_file(f.file, 0, upeer_sockaddr, | |
09952e3e JA |
1788 | upeer_addrlen, flags, |
1789 | rlimit(RLIMIT_NOFILE)); | |
6b07edeb | 1790 | fdput(f); |
de2ea4b6 JA |
1791 | } |
1792 | ||
1793 | return ret; | |
1da177e4 LT |
1794 | } |
1795 | ||
4541e805 DB |
1796 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1797 | int __user *, upeer_addrlen, int, flags) | |
1798 | { | |
1799 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, flags); | |
1800 | } | |
1801 | ||
20f37034 HC |
1802 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1803 | int __user *, upeer_addrlen) | |
aaca0bdc | 1804 | { |
4541e805 | 1805 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1806 | } |
1807 | ||
1da177e4 LT |
1808 | /* |
1809 | * Attempt to connect to a socket with the server address. The address | |
1810 | * is in user space so we verify it is OK and move it to kernel space. | |
1811 | * | |
1812 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1813 | * break bindings | |
1814 | * | |
1815 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1816 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1817 | * include the -EINPROGRESS status for such sockets. | |
1818 | */ | |
1819 | ||
f499a021 | 1820 | int __sys_connect_file(struct file *file, struct sockaddr_storage *address, |
bd3ded31 | 1821 | int addrlen, int file_flags) |
1da177e4 LT |
1822 | { |
1823 | struct socket *sock; | |
bd3ded31 | 1824 | int err; |
1da177e4 | 1825 | |
dba4a925 FR |
1826 | sock = sock_from_file(file); |
1827 | if (!sock) { | |
1828 | err = -ENOTSOCK; | |
1da177e4 | 1829 | goto out; |
dba4a925 | 1830 | } |
1da177e4 | 1831 | |
89bddce5 | 1832 | err = |
f499a021 | 1833 | security_socket_connect(sock, (struct sockaddr *)address, addrlen); |
1da177e4 | 1834 | if (err) |
bd3ded31 | 1835 | goto out; |
1da177e4 | 1836 | |
f499a021 | 1837 | err = sock->ops->connect(sock, (struct sockaddr *)address, addrlen, |
bd3ded31 | 1838 | sock->file->f_flags | file_flags); |
1da177e4 LT |
1839 | out: |
1840 | return err; | |
1841 | } | |
1842 | ||
bd3ded31 JA |
1843 | int __sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
1844 | { | |
1845 | int ret = -EBADF; | |
1846 | struct fd f; | |
1847 | ||
1848 | f = fdget(fd); | |
1849 | if (f.file) { | |
f499a021 JA |
1850 | struct sockaddr_storage address; |
1851 | ||
1852 | ret = move_addr_to_kernel(uservaddr, addrlen, &address); | |
1853 | if (!ret) | |
1854 | ret = __sys_connect_file(f.file, &address, addrlen, 0); | |
6b07edeb | 1855 | fdput(f); |
bd3ded31 JA |
1856 | } |
1857 | ||
1858 | return ret; | |
1859 | } | |
1860 | ||
1387c2c2 DB |
1861 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1862 | int, addrlen) | |
1863 | { | |
1864 | return __sys_connect(fd, uservaddr, addrlen); | |
1865 | } | |
1866 | ||
1da177e4 LT |
1867 | /* |
1868 | * Get the local address ('name') of a socket object. Move the obtained | |
1869 | * name to user space. | |
1870 | */ | |
1871 | ||
8882a107 DB |
1872 | int __sys_getsockname(int fd, struct sockaddr __user *usockaddr, |
1873 | int __user *usockaddr_len) | |
1da177e4 LT |
1874 | { |
1875 | struct socket *sock; | |
230b1839 | 1876 | struct sockaddr_storage address; |
9b2c45d4 | 1877 | int err, fput_needed; |
89bddce5 | 1878 | |
6cb153ca | 1879 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1880 | if (!sock) |
1881 | goto out; | |
1882 | ||
1883 | err = security_socket_getsockname(sock); | |
1884 | if (err) | |
1885 | goto out_put; | |
1886 | ||
9b2c45d4 DV |
1887 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 0); |
1888 | if (err < 0) | |
1da177e4 | 1889 | goto out_put; |
9b2c45d4 DV |
1890 | /* "err" is actually length in this case */ |
1891 | err = move_addr_to_user(&address, err, usockaddr, usockaddr_len); | |
1da177e4 LT |
1892 | |
1893 | out_put: | |
6cb153ca | 1894 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1895 | out: |
1896 | return err; | |
1897 | } | |
1898 | ||
8882a107 DB |
1899 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1900 | int __user *, usockaddr_len) | |
1901 | { | |
1902 | return __sys_getsockname(fd, usockaddr, usockaddr_len); | |
1903 | } | |
1904 | ||
1da177e4 LT |
1905 | /* |
1906 | * Get the remote address ('name') of a socket object. Move the obtained | |
1907 | * name to user space. | |
1908 | */ | |
1909 | ||
b21c8f83 DB |
1910 | int __sys_getpeername(int fd, struct sockaddr __user *usockaddr, |
1911 | int __user *usockaddr_len) | |
1da177e4 LT |
1912 | { |
1913 | struct socket *sock; | |
230b1839 | 1914 | struct sockaddr_storage address; |
9b2c45d4 | 1915 | int err, fput_needed; |
1da177e4 | 1916 | |
89bddce5 SH |
1917 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1918 | if (sock != NULL) { | |
1da177e4 LT |
1919 | err = security_socket_getpeername(sock); |
1920 | if (err) { | |
6cb153ca | 1921 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1922 | return err; |
1923 | } | |
1924 | ||
9b2c45d4 DV |
1925 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 1); |
1926 | if (err >= 0) | |
1927 | /* "err" is actually length in this case */ | |
1928 | err = move_addr_to_user(&address, err, usockaddr, | |
89bddce5 | 1929 | usockaddr_len); |
6cb153ca | 1930 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1931 | } |
1932 | return err; | |
1933 | } | |
1934 | ||
b21c8f83 DB |
1935 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1936 | int __user *, usockaddr_len) | |
1937 | { | |
1938 | return __sys_getpeername(fd, usockaddr, usockaddr_len); | |
1939 | } | |
1940 | ||
1da177e4 LT |
1941 | /* |
1942 | * Send a datagram to a given address. We move the address into kernel | |
1943 | * space and check the user space data area is readable before invoking | |
1944 | * the protocol. | |
1945 | */ | |
211b634b DB |
1946 | int __sys_sendto(int fd, void __user *buff, size_t len, unsigned int flags, |
1947 | struct sockaddr __user *addr, int addr_len) | |
1da177e4 LT |
1948 | { |
1949 | struct socket *sock; | |
230b1839 | 1950 | struct sockaddr_storage address; |
1da177e4 LT |
1951 | int err; |
1952 | struct msghdr msg; | |
1953 | struct iovec iov; | |
6cb153ca | 1954 | int fput_needed; |
6cb153ca | 1955 | |
602bd0e9 AV |
1956 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1957 | if (unlikely(err)) | |
1958 | return err; | |
de0fa95c PE |
1959 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1960 | if (!sock) | |
4387ff75 | 1961 | goto out; |
6cb153ca | 1962 | |
89bddce5 | 1963 | msg.msg_name = NULL; |
89bddce5 SH |
1964 | msg.msg_control = NULL; |
1965 | msg.msg_controllen = 0; | |
1966 | msg.msg_namelen = 0; | |
6cb153ca | 1967 | if (addr) { |
43db362d | 1968 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1969 | if (err < 0) |
1970 | goto out_put; | |
230b1839 | 1971 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1972 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1973 | } |
1974 | if (sock->file->f_flags & O_NONBLOCK) | |
1975 | flags |= MSG_DONTWAIT; | |
1976 | msg.msg_flags = flags; | |
d8725c86 | 1977 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 1978 | |
89bddce5 | 1979 | out_put: |
de0fa95c | 1980 | fput_light(sock->file, fput_needed); |
4387ff75 | 1981 | out: |
1da177e4 LT |
1982 | return err; |
1983 | } | |
1984 | ||
211b634b DB |
1985 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
1986 | unsigned int, flags, struct sockaddr __user *, addr, | |
1987 | int, addr_len) | |
1988 | { | |
1989 | return __sys_sendto(fd, buff, len, flags, addr, addr_len); | |
1990 | } | |
1991 | ||
1da177e4 | 1992 | /* |
89bddce5 | 1993 | * Send a datagram down a socket. |
1da177e4 LT |
1994 | */ |
1995 | ||
3e0fa65f | 1996 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1997 | unsigned int, flags) |
1da177e4 | 1998 | { |
211b634b | 1999 | return __sys_sendto(fd, buff, len, flags, NULL, 0); |
1da177e4 LT |
2000 | } |
2001 | ||
2002 | /* | |
89bddce5 | 2003 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
2004 | * sender. We verify the buffers are writable and if needed move the |
2005 | * sender address from kernel to user space. | |
2006 | */ | |
7a09e1eb DB |
2007 | int __sys_recvfrom(int fd, void __user *ubuf, size_t size, unsigned int flags, |
2008 | struct sockaddr __user *addr, int __user *addr_len) | |
1da177e4 LT |
2009 | { |
2010 | struct socket *sock; | |
2011 | struct iovec iov; | |
2012 | struct msghdr msg; | |
230b1839 | 2013 | struct sockaddr_storage address; |
89bddce5 | 2014 | int err, err2; |
6cb153ca BL |
2015 | int fput_needed; |
2016 | ||
602bd0e9 AV |
2017 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
2018 | if (unlikely(err)) | |
2019 | return err; | |
de0fa95c | 2020 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 2021 | if (!sock) |
de0fa95c | 2022 | goto out; |
1da177e4 | 2023 | |
89bddce5 SH |
2024 | msg.msg_control = NULL; |
2025 | msg.msg_controllen = 0; | |
f3d33426 HFS |
2026 | /* Save some cycles and don't copy the address if not needed */ |
2027 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
2028 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
2029 | msg.msg_namelen = 0; | |
130ed5d1 | 2030 | msg.msg_iocb = NULL; |
9f138fa6 | 2031 | msg.msg_flags = 0; |
1da177e4 LT |
2032 | if (sock->file->f_flags & O_NONBLOCK) |
2033 | flags |= MSG_DONTWAIT; | |
2da62906 | 2034 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 2035 | |
89bddce5 | 2036 | if (err >= 0 && addr != NULL) { |
43db362d | 2037 | err2 = move_addr_to_user(&address, |
230b1839 | 2038 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
2039 | if (err2 < 0) |
2040 | err = err2; | |
1da177e4 | 2041 | } |
de0fa95c PE |
2042 | |
2043 | fput_light(sock->file, fput_needed); | |
4387ff75 | 2044 | out: |
1da177e4 LT |
2045 | return err; |
2046 | } | |
2047 | ||
7a09e1eb DB |
2048 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
2049 | unsigned int, flags, struct sockaddr __user *, addr, | |
2050 | int __user *, addr_len) | |
2051 | { | |
2052 | return __sys_recvfrom(fd, ubuf, size, flags, addr, addr_len); | |
2053 | } | |
2054 | ||
1da177e4 | 2055 | /* |
89bddce5 | 2056 | * Receive a datagram from a socket. |
1da177e4 LT |
2057 | */ |
2058 | ||
b7c0ddf5 JG |
2059 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
2060 | unsigned int, flags) | |
1da177e4 | 2061 | { |
7a09e1eb | 2062 | return __sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
1da177e4 LT |
2063 | } |
2064 | ||
83f0c10b FW |
2065 | static bool sock_use_custom_sol_socket(const struct socket *sock) |
2066 | { | |
2067 | const struct sock *sk = sock->sk; | |
2068 | ||
2069 | /* Use sock->ops->setsockopt() for MPTCP */ | |
2070 | return IS_ENABLED(CONFIG_MPTCP) && | |
2071 | sk->sk_protocol == IPPROTO_MPTCP && | |
2072 | sk->sk_type == SOCK_STREAM && | |
2073 | (sk->sk_family == AF_INET || sk->sk_family == AF_INET6); | |
2074 | } | |
2075 | ||
1da177e4 LT |
2076 | /* |
2077 | * Set a socket option. Because we don't know the option lengths we have | |
2078 | * to pass the user mode parameter for the protocols to sort out. | |
2079 | */ | |
a7b75c5a | 2080 | int __sys_setsockopt(int fd, int level, int optname, char __user *user_optval, |
55db9c0e | 2081 | int optlen) |
1da177e4 | 2082 | { |
519a8a6c | 2083 | sockptr_t optval = USER_SOCKPTR(user_optval); |
0d01da6a | 2084 | char *kernel_optval = NULL; |
6cb153ca | 2085 | int err, fput_needed; |
1da177e4 LT |
2086 | struct socket *sock; |
2087 | ||
2088 | if (optlen < 0) | |
2089 | return -EINVAL; | |
89bddce5 SH |
2090 | |
2091 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
4a367299 CH |
2092 | if (!sock) |
2093 | return err; | |
1da177e4 | 2094 | |
4a367299 CH |
2095 | err = security_socket_setsockopt(sock, level, optname); |
2096 | if (err) | |
2097 | goto out_put; | |
0d01da6a | 2098 | |
55db9c0e CH |
2099 | if (!in_compat_syscall()) |
2100 | err = BPF_CGROUP_RUN_PROG_SETSOCKOPT(sock->sk, &level, &optname, | |
a7b75c5a | 2101 | user_optval, &optlen, |
55db9c0e | 2102 | &kernel_optval); |
4a367299 CH |
2103 | if (err < 0) |
2104 | goto out_put; | |
2105 | if (err > 0) { | |
2106 | err = 0; | |
2107 | goto out_put; | |
2108 | } | |
0d01da6a | 2109 | |
a7b75c5a CH |
2110 | if (kernel_optval) |
2111 | optval = KERNEL_SOCKPTR(kernel_optval); | |
4a367299 | 2112 | if (level == SOL_SOCKET && !sock_use_custom_sol_socket(sock)) |
a7b75c5a | 2113 | err = sock_setsockopt(sock, level, optname, optval, optlen); |
a44d9e72 CH |
2114 | else if (unlikely(!sock->ops->setsockopt)) |
2115 | err = -EOPNOTSUPP; | |
4a367299 CH |
2116 | else |
2117 | err = sock->ops->setsockopt(sock, level, optname, optval, | |
89bddce5 | 2118 | optlen); |
a7b75c5a | 2119 | kfree(kernel_optval); |
4a367299 CH |
2120 | out_put: |
2121 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2122 | return err; |
2123 | } | |
2124 | ||
cc36dca0 DB |
2125 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
2126 | char __user *, optval, int, optlen) | |
2127 | { | |
2128 | return __sys_setsockopt(fd, level, optname, optval, optlen); | |
2129 | } | |
2130 | ||
9cacf81f SF |
2131 | INDIRECT_CALLABLE_DECLARE(bool tcp_bpf_bypass_getsockopt(int level, |
2132 | int optname)); | |
2133 | ||
1da177e4 LT |
2134 | /* |
2135 | * Get a socket option. Because we don't know the option lengths we have | |
2136 | * to pass a user mode parameter for the protocols to sort out. | |
2137 | */ | |
55db9c0e CH |
2138 | int __sys_getsockopt(int fd, int level, int optname, char __user *optval, |
2139 | int __user *optlen) | |
1da177e4 | 2140 | { |
6cb153ca | 2141 | int err, fput_needed; |
1da177e4 | 2142 | struct socket *sock; |
0d01da6a | 2143 | int max_optlen; |
1da177e4 | 2144 | |
89bddce5 | 2145 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
d8a9b38f CH |
2146 | if (!sock) |
2147 | return err; | |
2148 | ||
2149 | err = security_socket_getsockopt(sock, level, optname); | |
2150 | if (err) | |
2151 | goto out_put; | |
1da177e4 | 2152 | |
55db9c0e CH |
2153 | if (!in_compat_syscall()) |
2154 | max_optlen = BPF_CGROUP_GETSOCKOPT_MAX_OPTLEN(optlen); | |
0d01da6a | 2155 | |
d8a9b38f CH |
2156 | if (level == SOL_SOCKET) |
2157 | err = sock_getsockopt(sock, level, optname, optval, optlen); | |
a44d9e72 CH |
2158 | else if (unlikely(!sock->ops->getsockopt)) |
2159 | err = -EOPNOTSUPP; | |
d8a9b38f CH |
2160 | else |
2161 | err = sock->ops->getsockopt(sock, level, optname, optval, | |
89bddce5 | 2162 | optlen); |
0d01da6a | 2163 | |
55db9c0e CH |
2164 | if (!in_compat_syscall()) |
2165 | err = BPF_CGROUP_RUN_PROG_GETSOCKOPT(sock->sk, level, optname, | |
2166 | optval, optlen, max_optlen, | |
2167 | err); | |
6cb153ca | 2168 | out_put: |
d8a9b38f | 2169 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2170 | return err; |
2171 | } | |
2172 | ||
13a2d70e DB |
2173 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
2174 | char __user *, optval, int __user *, optlen) | |
2175 | { | |
2176 | return __sys_getsockopt(fd, level, optname, optval, optlen); | |
2177 | } | |
2178 | ||
1da177e4 LT |
2179 | /* |
2180 | * Shutdown a socket. | |
2181 | */ | |
2182 | ||
b713c195 JA |
2183 | int __sys_shutdown_sock(struct socket *sock, int how) |
2184 | { | |
2185 | int err; | |
2186 | ||
2187 | err = security_socket_shutdown(sock, how); | |
2188 | if (!err) | |
2189 | err = sock->ops->shutdown(sock, how); | |
2190 | ||
2191 | return err; | |
2192 | } | |
2193 | ||
005a1aea | 2194 | int __sys_shutdown(int fd, int how) |
1da177e4 | 2195 | { |
6cb153ca | 2196 | int err, fput_needed; |
1da177e4 LT |
2197 | struct socket *sock; |
2198 | ||
89bddce5 SH |
2199 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2200 | if (sock != NULL) { | |
b713c195 | 2201 | err = __sys_shutdown_sock(sock, how); |
6cb153ca | 2202 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2203 | } |
2204 | return err; | |
2205 | } | |
2206 | ||
005a1aea DB |
2207 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
2208 | { | |
2209 | return __sys_shutdown(fd, how); | |
2210 | } | |
2211 | ||
89bddce5 | 2212 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
2213 | * fields which are the same type (int / unsigned) on our platforms. |
2214 | */ | |
2215 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
2216 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
2217 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
2218 | ||
c71d8ebe TH |
2219 | struct used_address { |
2220 | struct sockaddr_storage name; | |
2221 | unsigned int name_len; | |
2222 | }; | |
2223 | ||
0a384abf JA |
2224 | int __copy_msghdr_from_user(struct msghdr *kmsg, |
2225 | struct user_msghdr __user *umsg, | |
2226 | struct sockaddr __user **save_addr, | |
2227 | struct iovec __user **uiov, size_t *nsegs) | |
1661bf36 | 2228 | { |
ffb07550 | 2229 | struct user_msghdr msg; |
08adb7da AV |
2230 | ssize_t err; |
2231 | ||
ffb07550 | 2232 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 2233 | return -EFAULT; |
dbb490b9 | 2234 | |
1f466e1f CH |
2235 | kmsg->msg_control_is_user = true; |
2236 | kmsg->msg_control_user = msg.msg_control; | |
ffb07550 AV |
2237 | kmsg->msg_controllen = msg.msg_controllen; |
2238 | kmsg->msg_flags = msg.msg_flags; | |
2239 | ||
2240 | kmsg->msg_namelen = msg.msg_namelen; | |
2241 | if (!msg.msg_name) | |
6a2a2b3a AS |
2242 | kmsg->msg_namelen = 0; |
2243 | ||
dbb490b9 ML |
2244 | if (kmsg->msg_namelen < 0) |
2245 | return -EINVAL; | |
2246 | ||
1661bf36 | 2247 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 2248 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
2249 | |
2250 | if (save_addr) | |
ffb07550 | 2251 | *save_addr = msg.msg_name; |
08adb7da | 2252 | |
ffb07550 | 2253 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 2254 | if (!save_addr) { |
864d9664 PA |
2255 | err = move_addr_to_kernel(msg.msg_name, |
2256 | kmsg->msg_namelen, | |
08adb7da AV |
2257 | kmsg->msg_name); |
2258 | if (err < 0) | |
2259 | return err; | |
2260 | } | |
2261 | } else { | |
2262 | kmsg->msg_name = NULL; | |
2263 | kmsg->msg_namelen = 0; | |
2264 | } | |
2265 | ||
ffb07550 | 2266 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
2267 | return -EMSGSIZE; |
2268 | ||
0345f931 | 2269 | kmsg->msg_iocb = NULL; |
0a384abf JA |
2270 | *uiov = msg.msg_iov; |
2271 | *nsegs = msg.msg_iovlen; | |
2272 | return 0; | |
2273 | } | |
2274 | ||
2275 | static int copy_msghdr_from_user(struct msghdr *kmsg, | |
2276 | struct user_msghdr __user *umsg, | |
2277 | struct sockaddr __user **save_addr, | |
2278 | struct iovec **iov) | |
2279 | { | |
2280 | struct user_msghdr msg; | |
2281 | ssize_t err; | |
2282 | ||
2283 | err = __copy_msghdr_from_user(kmsg, umsg, save_addr, &msg.msg_iov, | |
2284 | &msg.msg_iovlen); | |
2285 | if (err) | |
2286 | return err; | |
0345f931 | 2287 | |
87e5e6da | 2288 | err = import_iovec(save_addr ? READ : WRITE, |
ffb07550 | 2289 | msg.msg_iov, msg.msg_iovlen, |
da184284 | 2290 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
87e5e6da | 2291 | return err < 0 ? err : 0; |
1661bf36 DC |
2292 | } |
2293 | ||
4257c8ca JA |
2294 | static int ____sys_sendmsg(struct socket *sock, struct msghdr *msg_sys, |
2295 | unsigned int flags, struct used_address *used_address, | |
2296 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 2297 | { |
b9d717a7 | 2298 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 2299 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 2300 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 2301 | unsigned char *ctl_buf = ctl; |
d8725c86 | 2302 | int ctl_len; |
08adb7da | 2303 | ssize_t err; |
89bddce5 | 2304 | |
1da177e4 LT |
2305 | err = -ENOBUFS; |
2306 | ||
228e548e | 2307 | if (msg_sys->msg_controllen > INT_MAX) |
4257c8ca | 2308 | goto out; |
28a94d8f | 2309 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2310 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2311 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2312 | err = |
228e548e | 2313 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2314 | sizeof(ctl)); |
1da177e4 | 2315 | if (err) |
4257c8ca | 2316 | goto out; |
228e548e AB |
2317 | ctl_buf = msg_sys->msg_control; |
2318 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2319 | } else if (ctl_len) { |
ac4340fc DM |
2320 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2321 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2322 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2323 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2324 | if (ctl_buf == NULL) |
4257c8ca | 2325 | goto out; |
1da177e4 LT |
2326 | } |
2327 | err = -EFAULT; | |
1f466e1f | 2328 | if (copy_from_user(ctl_buf, msg_sys->msg_control_user, ctl_len)) |
1da177e4 | 2329 | goto out_freectl; |
228e548e | 2330 | msg_sys->msg_control = ctl_buf; |
1f466e1f | 2331 | msg_sys->msg_control_is_user = false; |
1da177e4 | 2332 | } |
228e548e | 2333 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2334 | |
2335 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2336 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2337 | /* |
2338 | * If this is sendmmsg() and current destination address is same as | |
2339 | * previously succeeded address, omit asking LSM's decision. | |
2340 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2341 | * destination address never matches. | |
2342 | */ | |
bc909d9d MD |
2343 | if (used_address && msg_sys->msg_name && |
2344 | used_address->name_len == msg_sys->msg_namelen && | |
2345 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2346 | used_address->name_len)) { |
d8725c86 | 2347 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2348 | goto out_freectl; |
2349 | } | |
d8725c86 | 2350 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2351 | /* |
2352 | * If this is sendmmsg() and sending to current destination address was | |
2353 | * successful, remember it. | |
2354 | */ | |
2355 | if (used_address && err >= 0) { | |
2356 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2357 | if (msg_sys->msg_name) |
2358 | memcpy(&used_address->name, msg_sys->msg_name, | |
2359 | used_address->name_len); | |
c71d8ebe | 2360 | } |
1da177e4 LT |
2361 | |
2362 | out_freectl: | |
89bddce5 | 2363 | if (ctl_buf != ctl) |
1da177e4 | 2364 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
4257c8ca JA |
2365 | out: |
2366 | return err; | |
2367 | } | |
2368 | ||
03b1230c JA |
2369 | int sendmsg_copy_msghdr(struct msghdr *msg, |
2370 | struct user_msghdr __user *umsg, unsigned flags, | |
2371 | struct iovec **iov) | |
4257c8ca JA |
2372 | { |
2373 | int err; | |
2374 | ||
2375 | if (flags & MSG_CMSG_COMPAT) { | |
2376 | struct compat_msghdr __user *msg_compat; | |
2377 | ||
2378 | msg_compat = (struct compat_msghdr __user *) umsg; | |
2379 | err = get_compat_msghdr(msg, msg_compat, NULL, iov); | |
2380 | } else { | |
2381 | err = copy_msghdr_from_user(msg, umsg, NULL, iov); | |
2382 | } | |
2383 | if (err < 0) | |
2384 | return err; | |
2385 | ||
2386 | return 0; | |
2387 | } | |
2388 | ||
2389 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, | |
2390 | struct msghdr *msg_sys, unsigned int flags, | |
2391 | struct used_address *used_address, | |
2392 | unsigned int allowed_msghdr_flags) | |
2393 | { | |
2394 | struct sockaddr_storage address; | |
2395 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; | |
2396 | ssize_t err; | |
2397 | ||
2398 | msg_sys->msg_name = &address; | |
2399 | ||
2400 | err = sendmsg_copy_msghdr(msg_sys, msg, flags, &iov); | |
2401 | if (err < 0) | |
2402 | return err; | |
2403 | ||
2404 | err = ____sys_sendmsg(sock, msg_sys, flags, used_address, | |
2405 | allowed_msghdr_flags); | |
da184284 | 2406 | kfree(iov); |
228e548e AB |
2407 | return err; |
2408 | } | |
2409 | ||
2410 | /* | |
2411 | * BSD sendmsg interface | |
2412 | */ | |
03b1230c | 2413 | long __sys_sendmsg_sock(struct socket *sock, struct msghdr *msg, |
0fa03c62 JA |
2414 | unsigned int flags) |
2415 | { | |
03b1230c | 2416 | return ____sys_sendmsg(sock, msg, flags, NULL, 0); |
0fa03c62 | 2417 | } |
228e548e | 2418 | |
e1834a32 DB |
2419 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2420 | bool forbid_cmsg_compat) | |
228e548e AB |
2421 | { |
2422 | int fput_needed, err; | |
2423 | struct msghdr msg_sys; | |
1be374a0 AL |
2424 | struct socket *sock; |
2425 | ||
e1834a32 DB |
2426 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2427 | return -EINVAL; | |
2428 | ||
1be374a0 | 2429 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2430 | if (!sock) |
2431 | goto out; | |
2432 | ||
28a94d8f | 2433 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2434 | |
6cb153ca | 2435 | fput_light(sock->file, fput_needed); |
89bddce5 | 2436 | out: |
1da177e4 LT |
2437 | return err; |
2438 | } | |
2439 | ||
666547ff | 2440 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 | 2441 | { |
e1834a32 | 2442 | return __sys_sendmsg(fd, msg, flags, true); |
a7526eb5 AL |
2443 | } |
2444 | ||
228e548e AB |
2445 | /* |
2446 | * Linux sendmmsg interface | |
2447 | */ | |
2448 | ||
2449 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
e1834a32 | 2450 | unsigned int flags, bool forbid_cmsg_compat) |
228e548e AB |
2451 | { |
2452 | int fput_needed, err, datagrams; | |
2453 | struct socket *sock; | |
2454 | struct mmsghdr __user *entry; | |
2455 | struct compat_mmsghdr __user *compat_entry; | |
2456 | struct msghdr msg_sys; | |
c71d8ebe | 2457 | struct used_address used_address; |
f092276d | 2458 | unsigned int oflags = flags; |
228e548e | 2459 | |
e1834a32 DB |
2460 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2461 | return -EINVAL; | |
2462 | ||
98382f41 AB |
2463 | if (vlen > UIO_MAXIOV) |
2464 | vlen = UIO_MAXIOV; | |
228e548e AB |
2465 | |
2466 | datagrams = 0; | |
2467 | ||
2468 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2469 | if (!sock) | |
2470 | return err; | |
2471 | ||
c71d8ebe | 2472 | used_address.name_len = UINT_MAX; |
228e548e AB |
2473 | entry = mmsg; |
2474 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2475 | err = 0; |
f092276d | 2476 | flags |= MSG_BATCH; |
228e548e AB |
2477 | |
2478 | while (datagrams < vlen) { | |
f092276d TH |
2479 | if (datagrams == vlen - 1) |
2480 | flags = oflags; | |
2481 | ||
228e548e | 2482 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2483 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2484 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2485 | if (err < 0) |
2486 | break; | |
2487 | err = __put_user(err, &compat_entry->msg_len); | |
2488 | ++compat_entry; | |
2489 | } else { | |
a7526eb5 | 2490 | err = ___sys_sendmsg(sock, |
666547ff | 2491 | (struct user_msghdr __user *)entry, |
28a94d8f | 2492 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2493 | if (err < 0) |
2494 | break; | |
2495 | err = put_user(err, &entry->msg_len); | |
2496 | ++entry; | |
2497 | } | |
2498 | ||
2499 | if (err) | |
2500 | break; | |
2501 | ++datagrams; | |
3023898b SHY |
2502 | if (msg_data_left(&msg_sys)) |
2503 | break; | |
a78cb84c | 2504 | cond_resched(); |
228e548e AB |
2505 | } |
2506 | ||
228e548e AB |
2507 | fput_light(sock->file, fput_needed); |
2508 | ||
728ffb86 AB |
2509 | /* We only return an error if no datagrams were able to be sent */ |
2510 | if (datagrams != 0) | |
228e548e AB |
2511 | return datagrams; |
2512 | ||
228e548e AB |
2513 | return err; |
2514 | } | |
2515 | ||
2516 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2517 | unsigned int, vlen, unsigned int, flags) | |
2518 | { | |
e1834a32 | 2519 | return __sys_sendmmsg(fd, mmsg, vlen, flags, true); |
228e548e AB |
2520 | } |
2521 | ||
03b1230c JA |
2522 | int recvmsg_copy_msghdr(struct msghdr *msg, |
2523 | struct user_msghdr __user *umsg, unsigned flags, | |
2524 | struct sockaddr __user **uaddr, | |
2525 | struct iovec **iov) | |
1da177e4 | 2526 | { |
08adb7da | 2527 | ssize_t err; |
1da177e4 | 2528 | |
4257c8ca JA |
2529 | if (MSG_CMSG_COMPAT & flags) { |
2530 | struct compat_msghdr __user *msg_compat; | |
1da177e4 | 2531 | |
4257c8ca JA |
2532 | msg_compat = (struct compat_msghdr __user *) umsg; |
2533 | err = get_compat_msghdr(msg, msg_compat, uaddr, iov); | |
2534 | } else { | |
2535 | err = copy_msghdr_from_user(msg, umsg, uaddr, iov); | |
2536 | } | |
1da177e4 | 2537 | if (err < 0) |
da184284 | 2538 | return err; |
1da177e4 | 2539 | |
4257c8ca JA |
2540 | return 0; |
2541 | } | |
2542 | ||
2543 | static int ____sys_recvmsg(struct socket *sock, struct msghdr *msg_sys, | |
2544 | struct user_msghdr __user *msg, | |
2545 | struct sockaddr __user *uaddr, | |
2546 | unsigned int flags, int nosec) | |
2547 | { | |
2548 | struct compat_msghdr __user *msg_compat = | |
2549 | (struct compat_msghdr __user *) msg; | |
2550 | int __user *uaddr_len = COMPAT_NAMELEN(msg); | |
2551 | struct sockaddr_storage addr; | |
2552 | unsigned long cmsg_ptr; | |
2553 | int len; | |
2554 | ssize_t err; | |
2555 | ||
2556 | msg_sys->msg_name = &addr; | |
a2e27255 ACM |
2557 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2558 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2559 | |
f3d33426 HFS |
2560 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2561 | msg_sys->msg_namelen = 0; | |
2562 | ||
1da177e4 LT |
2563 | if (sock->file->f_flags & O_NONBLOCK) |
2564 | flags |= MSG_DONTWAIT; | |
1af66221 ED |
2565 | |
2566 | if (unlikely(nosec)) | |
2567 | err = sock_recvmsg_nosec(sock, msg_sys, flags); | |
2568 | else | |
2569 | err = sock_recvmsg(sock, msg_sys, flags); | |
2570 | ||
1da177e4 | 2571 | if (err < 0) |
4257c8ca | 2572 | goto out; |
1da177e4 LT |
2573 | len = err; |
2574 | ||
2575 | if (uaddr != NULL) { | |
43db362d | 2576 | err = move_addr_to_user(&addr, |
a2e27255 | 2577 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2578 | uaddr_len); |
1da177e4 | 2579 | if (err < 0) |
4257c8ca | 2580 | goto out; |
1da177e4 | 2581 | } |
a2e27255 | 2582 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2583 | COMPAT_FLAGS(msg)); |
1da177e4 | 2584 | if (err) |
4257c8ca | 2585 | goto out; |
1da177e4 | 2586 | if (MSG_CMSG_COMPAT & flags) |
a2e27255 | 2587 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2588 | &msg_compat->msg_controllen); |
2589 | else | |
a2e27255 | 2590 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2591 | &msg->msg_controllen); |
2592 | if (err) | |
4257c8ca | 2593 | goto out; |
1da177e4 | 2594 | err = len; |
4257c8ca JA |
2595 | out: |
2596 | return err; | |
2597 | } | |
2598 | ||
2599 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, | |
2600 | struct msghdr *msg_sys, unsigned int flags, int nosec) | |
2601 | { | |
2602 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; | |
2603 | /* user mode address pointers */ | |
2604 | struct sockaddr __user *uaddr; | |
2605 | ssize_t err; | |
2606 | ||
2607 | err = recvmsg_copy_msghdr(msg_sys, msg, flags, &uaddr, &iov); | |
2608 | if (err < 0) | |
2609 | return err; | |
1da177e4 | 2610 | |
4257c8ca | 2611 | err = ____sys_recvmsg(sock, msg_sys, msg, uaddr, flags, nosec); |
da184284 | 2612 | kfree(iov); |
a2e27255 ACM |
2613 | return err; |
2614 | } | |
2615 | ||
2616 | /* | |
2617 | * BSD recvmsg interface | |
2618 | */ | |
2619 | ||
03b1230c JA |
2620 | long __sys_recvmsg_sock(struct socket *sock, struct msghdr *msg, |
2621 | struct user_msghdr __user *umsg, | |
2622 | struct sockaddr __user *uaddr, unsigned int flags) | |
aa1fa28f | 2623 | { |
03b1230c | 2624 | return ____sys_recvmsg(sock, msg, umsg, uaddr, flags, 0); |
aa1fa28f JA |
2625 | } |
2626 | ||
e1834a32 DB |
2627 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2628 | bool forbid_cmsg_compat) | |
a2e27255 ACM |
2629 | { |
2630 | int fput_needed, err; | |
2631 | struct msghdr msg_sys; | |
1be374a0 AL |
2632 | struct socket *sock; |
2633 | ||
e1834a32 DB |
2634 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2635 | return -EINVAL; | |
2636 | ||
1be374a0 | 2637 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2638 | if (!sock) |
2639 | goto out; | |
2640 | ||
a7526eb5 | 2641 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2642 | |
6cb153ca | 2643 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2644 | out: |
2645 | return err; | |
2646 | } | |
2647 | ||
666547ff | 2648 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2649 | unsigned int, flags) |
2650 | { | |
e1834a32 | 2651 | return __sys_recvmsg(fd, msg, flags, true); |
a7526eb5 AL |
2652 | } |
2653 | ||
a2e27255 ACM |
2654 | /* |
2655 | * Linux recvmmsg interface | |
2656 | */ | |
2657 | ||
e11d4284 AB |
2658 | static int do_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2659 | unsigned int vlen, unsigned int flags, | |
2660 | struct timespec64 *timeout) | |
a2e27255 ACM |
2661 | { |
2662 | int fput_needed, err, datagrams; | |
2663 | struct socket *sock; | |
2664 | struct mmsghdr __user *entry; | |
d7256d0e | 2665 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2666 | struct msghdr msg_sys; |
766b9f92 DD |
2667 | struct timespec64 end_time; |
2668 | struct timespec64 timeout64; | |
a2e27255 ACM |
2669 | |
2670 | if (timeout && | |
2671 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2672 | timeout->tv_nsec)) | |
2673 | return -EINVAL; | |
2674 | ||
2675 | datagrams = 0; | |
2676 | ||
2677 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2678 | if (!sock) | |
2679 | return err; | |
2680 | ||
7797dc41 SHY |
2681 | if (likely(!(flags & MSG_ERRQUEUE))) { |
2682 | err = sock_error(sock->sk); | |
2683 | if (err) { | |
2684 | datagrams = err; | |
2685 | goto out_put; | |
2686 | } | |
e623a9e9 | 2687 | } |
a2e27255 ACM |
2688 | |
2689 | entry = mmsg; | |
d7256d0e | 2690 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2691 | |
2692 | while (datagrams < vlen) { | |
2693 | /* | |
2694 | * No need to ask LSM for more than the first datagram. | |
2695 | */ | |
d7256d0e | 2696 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2697 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2698 | &msg_sys, flags & ~MSG_WAITFORONE, |
2699 | datagrams); | |
d7256d0e JMG |
2700 | if (err < 0) |
2701 | break; | |
2702 | err = __put_user(err, &compat_entry->msg_len); | |
2703 | ++compat_entry; | |
2704 | } else { | |
a7526eb5 | 2705 | err = ___sys_recvmsg(sock, |
666547ff | 2706 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2707 | &msg_sys, flags & ~MSG_WAITFORONE, |
2708 | datagrams); | |
d7256d0e JMG |
2709 | if (err < 0) |
2710 | break; | |
2711 | err = put_user(err, &entry->msg_len); | |
2712 | ++entry; | |
2713 | } | |
2714 | ||
a2e27255 ACM |
2715 | if (err) |
2716 | break; | |
a2e27255 ACM |
2717 | ++datagrams; |
2718 | ||
71c5c159 BB |
2719 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2720 | if (flags & MSG_WAITFORONE) | |
2721 | flags |= MSG_DONTWAIT; | |
2722 | ||
a2e27255 | 2723 | if (timeout) { |
766b9f92 | 2724 | ktime_get_ts64(&timeout64); |
c2e6c856 | 2725 | *timeout = timespec64_sub(end_time, timeout64); |
a2e27255 ACM |
2726 | if (timeout->tv_sec < 0) { |
2727 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2728 | break; | |
2729 | } | |
2730 | ||
2731 | /* Timeout, return less than vlen datagrams */ | |
2732 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2733 | break; | |
2734 | } | |
2735 | ||
2736 | /* Out of band data, return right away */ | |
2737 | if (msg_sys.msg_flags & MSG_OOB) | |
2738 | break; | |
a78cb84c | 2739 | cond_resched(); |
a2e27255 ACM |
2740 | } |
2741 | ||
a2e27255 | 2742 | if (err == 0) |
34b88a68 ACM |
2743 | goto out_put; |
2744 | ||
2745 | if (datagrams == 0) { | |
2746 | datagrams = err; | |
2747 | goto out_put; | |
2748 | } | |
a2e27255 | 2749 | |
34b88a68 ACM |
2750 | /* |
2751 | * We may return less entries than requested (vlen) if the | |
2752 | * sock is non block and there aren't enough datagrams... | |
2753 | */ | |
2754 | if (err != -EAGAIN) { | |
a2e27255 | 2755 | /* |
34b88a68 ACM |
2756 | * ... or if recvmsg returns an error after we |
2757 | * received some datagrams, where we record the | |
2758 | * error to return on the next call or if the | |
2759 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2760 | */ |
34b88a68 | 2761 | sock->sk->sk_err = -err; |
a2e27255 | 2762 | } |
34b88a68 ACM |
2763 | out_put: |
2764 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2765 | |
34b88a68 | 2766 | return datagrams; |
a2e27255 ACM |
2767 | } |
2768 | ||
e11d4284 AB |
2769 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2770 | unsigned int vlen, unsigned int flags, | |
2771 | struct __kernel_timespec __user *timeout, | |
2772 | struct old_timespec32 __user *timeout32) | |
a2e27255 ACM |
2773 | { |
2774 | int datagrams; | |
c2e6c856 | 2775 | struct timespec64 timeout_sys; |
a2e27255 | 2776 | |
e11d4284 AB |
2777 | if (timeout && get_timespec64(&timeout_sys, timeout)) |
2778 | return -EFAULT; | |
a2e27255 | 2779 | |
e11d4284 | 2780 | if (timeout32 && get_old_timespec32(&timeout_sys, timeout32)) |
a2e27255 ACM |
2781 | return -EFAULT; |
2782 | ||
e11d4284 AB |
2783 | if (!timeout && !timeout32) |
2784 | return do_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2785 | ||
2786 | datagrams = do_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
a2e27255 | 2787 | |
e11d4284 AB |
2788 | if (datagrams <= 0) |
2789 | return datagrams; | |
2790 | ||
2791 | if (timeout && put_timespec64(&timeout_sys, timeout)) | |
2792 | datagrams = -EFAULT; | |
2793 | ||
2794 | if (timeout32 && put_old_timespec32(&timeout_sys, timeout32)) | |
a2e27255 ACM |
2795 | datagrams = -EFAULT; |
2796 | ||
2797 | return datagrams; | |
2798 | } | |
2799 | ||
1255e269 DB |
2800 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, |
2801 | unsigned int, vlen, unsigned int, flags, | |
c2e6c856 | 2802 | struct __kernel_timespec __user *, timeout) |
1255e269 | 2803 | { |
e11d4284 AB |
2804 | if (flags & MSG_CMSG_COMPAT) |
2805 | return -EINVAL; | |
2806 | ||
2807 | return __sys_recvmmsg(fd, mmsg, vlen, flags, timeout, NULL); | |
2808 | } | |
2809 | ||
2810 | #ifdef CONFIG_COMPAT_32BIT_TIME | |
2811 | SYSCALL_DEFINE5(recvmmsg_time32, int, fd, struct mmsghdr __user *, mmsg, | |
2812 | unsigned int, vlen, unsigned int, flags, | |
2813 | struct old_timespec32 __user *, timeout) | |
2814 | { | |
2815 | if (flags & MSG_CMSG_COMPAT) | |
2816 | return -EINVAL; | |
2817 | ||
2818 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL, timeout); | |
1255e269 | 2819 | } |
e11d4284 | 2820 | #endif |
1255e269 | 2821 | |
a2e27255 | 2822 | #ifdef __ARCH_WANT_SYS_SOCKETCALL |
1da177e4 LT |
2823 | /* Argument list sizes for sys_socketcall */ |
2824 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2825 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2826 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2827 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2828 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2829 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2830 | }; |
2831 | ||
1da177e4 LT |
2832 | #undef AL |
2833 | ||
2834 | /* | |
89bddce5 | 2835 | * System call vectors. |
1da177e4 LT |
2836 | * |
2837 | * Argument checking cleaned up. Saved 20% in size. | |
2838 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2839 | * it is set by the callees. |
1da177e4 LT |
2840 | */ |
2841 | ||
3e0fa65f | 2842 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2843 | { |
2950fa9d | 2844 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2845 | unsigned long a0, a1; |
1da177e4 | 2846 | int err; |
47379052 | 2847 | unsigned int len; |
1da177e4 | 2848 | |
228e548e | 2849 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 | 2850 | return -EINVAL; |
c8e8cd57 | 2851 | call = array_index_nospec(call, SYS_SENDMMSG + 1); |
1da177e4 | 2852 | |
47379052 AV |
2853 | len = nargs[call]; |
2854 | if (len > sizeof(a)) | |
2855 | return -EINVAL; | |
2856 | ||
1da177e4 | 2857 | /* copy_from_user should be SMP safe. */ |
47379052 | 2858 | if (copy_from_user(a, args, len)) |
1da177e4 | 2859 | return -EFAULT; |
3ec3b2fb | 2860 | |
2950fa9d CG |
2861 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2862 | if (err) | |
2863 | return err; | |
3ec3b2fb | 2864 | |
89bddce5 SH |
2865 | a0 = a[0]; |
2866 | a1 = a[1]; | |
2867 | ||
2868 | switch (call) { | |
2869 | case SYS_SOCKET: | |
9d6a15c3 | 2870 | err = __sys_socket(a0, a1, a[2]); |
89bddce5 SH |
2871 | break; |
2872 | case SYS_BIND: | |
a87d35d8 | 2873 | err = __sys_bind(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2874 | break; |
2875 | case SYS_CONNECT: | |
1387c2c2 | 2876 | err = __sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2877 | break; |
2878 | case SYS_LISTEN: | |
25e290ee | 2879 | err = __sys_listen(a0, a1); |
89bddce5 SH |
2880 | break; |
2881 | case SYS_ACCEPT: | |
4541e805 DB |
2882 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2883 | (int __user *)a[2], 0); | |
89bddce5 SH |
2884 | break; |
2885 | case SYS_GETSOCKNAME: | |
2886 | err = | |
8882a107 DB |
2887 | __sys_getsockname(a0, (struct sockaddr __user *)a1, |
2888 | (int __user *)a[2]); | |
89bddce5 SH |
2889 | break; |
2890 | case SYS_GETPEERNAME: | |
2891 | err = | |
b21c8f83 DB |
2892 | __sys_getpeername(a0, (struct sockaddr __user *)a1, |
2893 | (int __user *)a[2]); | |
89bddce5 SH |
2894 | break; |
2895 | case SYS_SOCKETPAIR: | |
6debc8d8 | 2896 | err = __sys_socketpair(a0, a1, a[2], (int __user *)a[3]); |
89bddce5 SH |
2897 | break; |
2898 | case SYS_SEND: | |
f3bf896b DB |
2899 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2900 | NULL, 0); | |
89bddce5 SH |
2901 | break; |
2902 | case SYS_SENDTO: | |
211b634b DB |
2903 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2904 | (struct sockaddr __user *)a[4], a[5]); | |
89bddce5 SH |
2905 | break; |
2906 | case SYS_RECV: | |
d27e9afc DB |
2907 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2908 | NULL, NULL); | |
89bddce5 SH |
2909 | break; |
2910 | case SYS_RECVFROM: | |
7a09e1eb DB |
2911 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2912 | (struct sockaddr __user *)a[4], | |
2913 | (int __user *)a[5]); | |
89bddce5 SH |
2914 | break; |
2915 | case SYS_SHUTDOWN: | |
005a1aea | 2916 | err = __sys_shutdown(a0, a1); |
89bddce5 SH |
2917 | break; |
2918 | case SYS_SETSOCKOPT: | |
cc36dca0 DB |
2919 | err = __sys_setsockopt(a0, a1, a[2], (char __user *)a[3], |
2920 | a[4]); | |
89bddce5 SH |
2921 | break; |
2922 | case SYS_GETSOCKOPT: | |
2923 | err = | |
13a2d70e DB |
2924 | __sys_getsockopt(a0, a1, a[2], (char __user *)a[3], |
2925 | (int __user *)a[4]); | |
89bddce5 SH |
2926 | break; |
2927 | case SYS_SENDMSG: | |
e1834a32 DB |
2928 | err = __sys_sendmsg(a0, (struct user_msghdr __user *)a1, |
2929 | a[2], true); | |
89bddce5 | 2930 | break; |
228e548e | 2931 | case SYS_SENDMMSG: |
e1834a32 DB |
2932 | err = __sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
2933 | a[3], true); | |
228e548e | 2934 | break; |
89bddce5 | 2935 | case SYS_RECVMSG: |
e1834a32 DB |
2936 | err = __sys_recvmsg(a0, (struct user_msghdr __user *)a1, |
2937 | a[2], true); | |
89bddce5 | 2938 | break; |
a2e27255 | 2939 | case SYS_RECVMMSG: |
3ca47e95 | 2940 | if (IS_ENABLED(CONFIG_64BIT)) |
e11d4284 AB |
2941 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, |
2942 | a[2], a[3], | |
2943 | (struct __kernel_timespec __user *)a[4], | |
2944 | NULL); | |
2945 | else | |
2946 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, | |
2947 | a[2], a[3], NULL, | |
2948 | (struct old_timespec32 __user *)a[4]); | |
a2e27255 | 2949 | break; |
de11defe | 2950 | case SYS_ACCEPT4: |
4541e805 DB |
2951 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2952 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2953 | break; |
89bddce5 SH |
2954 | default: |
2955 | err = -EINVAL; | |
2956 | break; | |
1da177e4 LT |
2957 | } |
2958 | return err; | |
2959 | } | |
2960 | ||
89bddce5 | 2961 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2962 | |
55737fda SH |
2963 | /** |
2964 | * sock_register - add a socket protocol handler | |
2965 | * @ops: description of protocol | |
2966 | * | |
1da177e4 LT |
2967 | * This function is called by a protocol handler that wants to |
2968 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2969 | * socket interface. The value ops->family corresponds to the |
55737fda | 2970 | * socket system call protocol family. |
1da177e4 | 2971 | */ |
f0fd27d4 | 2972 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2973 | { |
2974 | int err; | |
2975 | ||
2976 | if (ops->family >= NPROTO) { | |
3410f22e | 2977 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2978 | return -ENOBUFS; |
2979 | } | |
55737fda SH |
2980 | |
2981 | spin_lock(&net_family_lock); | |
190683a9 ED |
2982 | if (rcu_dereference_protected(net_families[ops->family], |
2983 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2984 | err = -EEXIST; |
2985 | else { | |
cf778b00 | 2986 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2987 | err = 0; |
2988 | } | |
55737fda SH |
2989 | spin_unlock(&net_family_lock); |
2990 | ||
3410f22e | 2991 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2992 | return err; |
2993 | } | |
c6d409cf | 2994 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2995 | |
55737fda SH |
2996 | /** |
2997 | * sock_unregister - remove a protocol handler | |
2998 | * @family: protocol family to remove | |
2999 | * | |
1da177e4 LT |
3000 | * This function is called by a protocol handler that wants to |
3001 | * remove its address family, and have it unlinked from the | |
55737fda SH |
3002 | * new socket creation. |
3003 | * | |
3004 | * If protocol handler is a module, then it can use module reference | |
3005 | * counts to protect against new references. If protocol handler is not | |
3006 | * a module then it needs to provide its own protection in | |
3007 | * the ops->create routine. | |
1da177e4 | 3008 | */ |
f0fd27d4 | 3009 | void sock_unregister(int family) |
1da177e4 | 3010 | { |
f0fd27d4 | 3011 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 3012 | |
55737fda | 3013 | spin_lock(&net_family_lock); |
a9b3cd7f | 3014 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
3015 | spin_unlock(&net_family_lock); |
3016 | ||
3017 | synchronize_rcu(); | |
3018 | ||
3410f22e | 3019 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 3020 | } |
c6d409cf | 3021 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 3022 | |
bf2ae2e4 XL |
3023 | bool sock_is_registered(int family) |
3024 | { | |
66b51b0a | 3025 | return family < NPROTO && rcu_access_pointer(net_families[family]); |
bf2ae2e4 XL |
3026 | } |
3027 | ||
77d76ea3 | 3028 | static int __init sock_init(void) |
1da177e4 | 3029 | { |
b3e19d92 | 3030 | int err; |
2ca794e5 EB |
3031 | /* |
3032 | * Initialize the network sysctl infrastructure. | |
3033 | */ | |
3034 | err = net_sysctl_init(); | |
3035 | if (err) | |
3036 | goto out; | |
b3e19d92 | 3037 | |
1da177e4 | 3038 | /* |
89bddce5 | 3039 | * Initialize skbuff SLAB cache |
1da177e4 LT |
3040 | */ |
3041 | skb_init(); | |
1da177e4 LT |
3042 | |
3043 | /* | |
89bddce5 | 3044 | * Initialize the protocols module. |
1da177e4 LT |
3045 | */ |
3046 | ||
3047 | init_inodecache(); | |
b3e19d92 NP |
3048 | |
3049 | err = register_filesystem(&sock_fs_type); | |
3050 | if (err) | |
47260ba9 | 3051 | goto out; |
1da177e4 | 3052 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
3053 | if (IS_ERR(sock_mnt)) { |
3054 | err = PTR_ERR(sock_mnt); | |
3055 | goto out_mount; | |
3056 | } | |
77d76ea3 AK |
3057 | |
3058 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
3059 | */ |
3060 | ||
3061 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
3062 | err = netfilter_init(); |
3063 | if (err) | |
3064 | goto out; | |
1da177e4 | 3065 | #endif |
cbeb321a | 3066 | |
408eccce | 3067 | ptp_classifier_init(); |
c1f19b51 | 3068 | |
b3e19d92 NP |
3069 | out: |
3070 | return err; | |
3071 | ||
3072 | out_mount: | |
3073 | unregister_filesystem(&sock_fs_type); | |
b3e19d92 | 3074 | goto out; |
1da177e4 LT |
3075 | } |
3076 | ||
77d76ea3 AK |
3077 | core_initcall(sock_init); /* early initcall */ |
3078 | ||
1da177e4 LT |
3079 | #ifdef CONFIG_PROC_FS |
3080 | void socket_seq_show(struct seq_file *seq) | |
3081 | { | |
648845ab TZ |
3082 | seq_printf(seq, "sockets: used %d\n", |
3083 | sock_inuse_get(seq->private)); | |
1da177e4 | 3084 | } |
89bddce5 | 3085 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 3086 | |
89bbfc95 | 3087 | #ifdef CONFIG_COMPAT |
36fd633e | 3088 | static int compat_dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 3089 | { |
6b96018b | 3090 | struct compat_ifconf ifc32; |
7a229387 | 3091 | struct ifconf ifc; |
7a229387 AB |
3092 | int err; |
3093 | ||
6b96018b | 3094 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
3095 | return -EFAULT; |
3096 | ||
36fd633e AV |
3097 | ifc.ifc_len = ifc32.ifc_len; |
3098 | ifc.ifc_req = compat_ptr(ifc32.ifcbuf); | |
7a229387 | 3099 | |
36fd633e AV |
3100 | rtnl_lock(); |
3101 | err = dev_ifconf(net, &ifc, sizeof(struct compat_ifreq)); | |
3102 | rtnl_unlock(); | |
7a229387 AB |
3103 | if (err) |
3104 | return err; | |
3105 | ||
36fd633e | 3106 | ifc32.ifc_len = ifc.ifc_len; |
6b96018b | 3107 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
3108 | return -EFAULT; |
3109 | ||
3110 | return 0; | |
3111 | } | |
3112 | ||
6b96018b | 3113 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 3114 | { |
3a7da39d BH |
3115 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
3116 | bool convert_in = false, convert_out = false; | |
44c02a2c AV |
3117 | size_t buf_size = 0; |
3118 | struct ethtool_rxnfc __user *rxnfc = NULL; | |
3119 | struct ifreq ifr; | |
3a7da39d BH |
3120 | u32 rule_cnt = 0, actual_rule_cnt; |
3121 | u32 ethcmd; | |
7a229387 | 3122 | u32 data; |
3a7da39d | 3123 | int ret; |
7a229387 | 3124 | |
3a7da39d BH |
3125 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
3126 | return -EFAULT; | |
7a229387 | 3127 | |
3a7da39d BH |
3128 | compat_rxnfc = compat_ptr(data); |
3129 | ||
3130 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
3131 | return -EFAULT; |
3132 | ||
3a7da39d BH |
3133 | /* Most ethtool structures are defined without padding. |
3134 | * Unfortunately struct ethtool_rxnfc is an exception. | |
3135 | */ | |
3136 | switch (ethcmd) { | |
3137 | default: | |
3138 | break; | |
3139 | case ETHTOOL_GRXCLSRLALL: | |
3140 | /* Buffer size is variable */ | |
3141 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
3142 | return -EFAULT; | |
3143 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
3144 | return -ENOMEM; | |
3145 | buf_size += rule_cnt * sizeof(u32); | |
7c7ab580 | 3146 | fallthrough; |
3a7da39d BH |
3147 | case ETHTOOL_GRXRINGS: |
3148 | case ETHTOOL_GRXCLSRLCNT: | |
3149 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 3150 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d | 3151 | convert_out = true; |
7c7ab580 | 3152 | fallthrough; |
3a7da39d | 3153 | case ETHTOOL_SRXCLSRLDEL: |
3a7da39d BH |
3154 | buf_size += sizeof(struct ethtool_rxnfc); |
3155 | convert_in = true; | |
44c02a2c | 3156 | rxnfc = compat_alloc_user_space(buf_size); |
3a7da39d BH |
3157 | break; |
3158 | } | |
3159 | ||
44c02a2c | 3160 | if (copy_from_user(&ifr.ifr_name, &ifr32->ifr_name, IFNAMSIZ)) |
7a229387 AB |
3161 | return -EFAULT; |
3162 | ||
44c02a2c | 3163 | ifr.ifr_data = convert_in ? rxnfc : (void __user *)compat_rxnfc; |
7a229387 | 3164 | |
3a7da39d | 3165 | if (convert_in) { |
127fe533 | 3166 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
3167 | * fs.ring_cookie and at the end of fs, but nowhere else. |
3168 | */ | |
127fe533 AD |
3169 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
3170 | sizeof(compat_rxnfc->fs.m_ext) != | |
3171 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
3172 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
3173 | BUILD_BUG_ON( |
3174 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
3175 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
3176 | offsetof(struct ethtool_rxnfc, fs.location) - | |
3177 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
3178 | ||
3179 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
3180 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
3181 | (void __user *)rxnfc) || | |
3a7da39d BH |
3182 | copy_in_user(&rxnfc->fs.ring_cookie, |
3183 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 | 3184 | (void __user *)(&rxnfc->fs.location + 1) - |
b6168562 WW |
3185 | (void __user *)&rxnfc->fs.ring_cookie)) |
3186 | return -EFAULT; | |
3187 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3188 | if (put_user(rule_cnt, &rxnfc->rule_cnt)) | |
3189 | return -EFAULT; | |
3190 | } else if (copy_in_user(&rxnfc->rule_cnt, | |
3191 | &compat_rxnfc->rule_cnt, | |
3192 | sizeof(rxnfc->rule_cnt))) | |
3a7da39d BH |
3193 | return -EFAULT; |
3194 | } | |
3195 | ||
44c02a2c | 3196 | ret = dev_ioctl(net, SIOCETHTOOL, &ifr, NULL); |
3a7da39d BH |
3197 | if (ret) |
3198 | return ret; | |
3199 | ||
3200 | if (convert_out) { | |
3201 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
3202 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
3203 | (const void __user *)rxnfc) || | |
3a7da39d BH |
3204 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
3205 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
3206 | (const void __user *)(&rxnfc->fs.location + 1) - |
3207 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
3208 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
3209 | sizeof(rxnfc->rule_cnt))) | |
3210 | return -EFAULT; | |
3211 | ||
3212 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
3213 | /* As an optimisation, we only copy the actual | |
3214 | * number of rules that the underlying | |
3215 | * function returned. Since Mallory might | |
3216 | * change the rule count in user memory, we | |
3217 | * check that it is less than the rule count | |
3218 | * originally given (as the user buffer size), | |
3219 | * which has been range-checked. | |
3220 | */ | |
3221 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
3222 | return -EFAULT; | |
3223 | if (actual_rule_cnt < rule_cnt) | |
3224 | rule_cnt = actual_rule_cnt; | |
3225 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
3226 | &rxnfc->rule_locs[0], | |
3227 | rule_cnt * sizeof(u32))) | |
3228 | return -EFAULT; | |
3229 | } | |
3230 | } | |
3231 | ||
3232 | return 0; | |
7a229387 AB |
3233 | } |
3234 | ||
7a50a240 AB |
3235 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
3236 | { | |
7a50a240 | 3237 | compat_uptr_t uptr32; |
44c02a2c AV |
3238 | struct ifreq ifr; |
3239 | void __user *saved; | |
3240 | int err; | |
7a50a240 | 3241 | |
44c02a2c | 3242 | if (copy_from_user(&ifr, uifr32, sizeof(struct compat_ifreq))) |
7a50a240 AB |
3243 | return -EFAULT; |
3244 | ||
3245 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
3246 | return -EFAULT; | |
3247 | ||
44c02a2c AV |
3248 | saved = ifr.ifr_settings.ifs_ifsu.raw_hdlc; |
3249 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = compat_ptr(uptr32); | |
7a229387 | 3250 | |
44c02a2c AV |
3251 | err = dev_ioctl(net, SIOCWANDEV, &ifr, NULL); |
3252 | if (!err) { | |
3253 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = saved; | |
3254 | if (copy_to_user(uifr32, &ifr, sizeof(struct compat_ifreq))) | |
3255 | err = -EFAULT; | |
ccbd6a5a | 3256 | } |
44c02a2c | 3257 | return err; |
7a229387 AB |
3258 | } |
3259 | ||
590d4693 BH |
3260 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
3261 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 3262 | struct compat_ifreq __user *u_ifreq32) |
7a229387 | 3263 | { |
44c02a2c | 3264 | struct ifreq ifreq; |
7a229387 AB |
3265 | u32 data32; |
3266 | ||
44c02a2c | 3267 | if (copy_from_user(ifreq.ifr_name, u_ifreq32->ifr_name, IFNAMSIZ)) |
7a229387 | 3268 | return -EFAULT; |
44c02a2c | 3269 | if (get_user(data32, &u_ifreq32->ifr_data)) |
7a229387 | 3270 | return -EFAULT; |
44c02a2c | 3271 | ifreq.ifr_data = compat_ptr(data32); |
7a229387 | 3272 | |
44c02a2c | 3273 | return dev_ioctl(net, cmd, &ifreq, NULL); |
7a229387 AB |
3274 | } |
3275 | ||
37ac39bd JB |
3276 | static int compat_ifreq_ioctl(struct net *net, struct socket *sock, |
3277 | unsigned int cmd, | |
3278 | struct compat_ifreq __user *uifr32) | |
3279 | { | |
3280 | struct ifreq __user *uifr; | |
3281 | int err; | |
3282 | ||
3283 | /* Handle the fact that while struct ifreq has the same *layout* on | |
3284 | * 32/64 for everything but ifreq::ifru_ifmap and ifreq::ifru_data, | |
3285 | * which are handled elsewhere, it still has different *size* due to | |
3286 | * ifreq::ifru_ifmap (which is 16 bytes on 32 bit, 24 bytes on 64-bit, | |
3287 | * resulting in struct ifreq being 32 and 40 bytes respectively). | |
3288 | * As a result, if the struct happens to be at the end of a page and | |
3289 | * the next page isn't readable/writable, we get a fault. To prevent | |
3290 | * that, copy back and forth to the full size. | |
3291 | */ | |
3292 | ||
3293 | uifr = compat_alloc_user_space(sizeof(*uifr)); | |
3294 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
3295 | return -EFAULT; | |
3296 | ||
3297 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
3298 | ||
3299 | if (!err) { | |
3300 | switch (cmd) { | |
3301 | case SIOCGIFFLAGS: | |
3302 | case SIOCGIFMETRIC: | |
3303 | case SIOCGIFMTU: | |
3304 | case SIOCGIFMEM: | |
3305 | case SIOCGIFHWADDR: | |
3306 | case SIOCGIFINDEX: | |
3307 | case SIOCGIFADDR: | |
3308 | case SIOCGIFBRDADDR: | |
3309 | case SIOCGIFDSTADDR: | |
3310 | case SIOCGIFNETMASK: | |
3311 | case SIOCGIFPFLAGS: | |
3312 | case SIOCGIFTXQLEN: | |
3313 | case SIOCGMIIPHY: | |
3314 | case SIOCGMIIREG: | |
c6c9fee3 | 3315 | case SIOCGIFNAME: |
37ac39bd JB |
3316 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
3317 | err = -EFAULT; | |
3318 | break; | |
3319 | } | |
3320 | } | |
3321 | return err; | |
3322 | } | |
3323 | ||
a2116ed2 AB |
3324 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
3325 | struct compat_ifreq __user *uifr32) | |
3326 | { | |
3327 | struct ifreq ifr; | |
3328 | struct compat_ifmap __user *uifmap32; | |
a2116ed2 AB |
3329 | int err; |
3330 | ||
3331 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
3332 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3333 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3334 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3335 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3336 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3337 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3338 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3339 | if (err) |
3340 | return -EFAULT; | |
3341 | ||
44c02a2c | 3342 | err = dev_ioctl(net, cmd, &ifr, NULL); |
a2116ed2 AB |
3343 | |
3344 | if (cmd == SIOCGIFMAP && !err) { | |
3345 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3346 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3347 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3348 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3349 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3350 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3351 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3352 | if (err) |
3353 | err = -EFAULT; | |
3354 | } | |
3355 | return err; | |
3356 | } | |
3357 | ||
7a229387 AB |
3358 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE |
3359 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3360 | * use compatible ioctls |
7a229387 | 3361 | */ |
6b96018b | 3362 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3363 | { |
6b96018b | 3364 | compat_ulong_t tmp; |
7a229387 | 3365 | |
6b96018b | 3366 | if (get_user(tmp, argp)) |
7a229387 AB |
3367 | return -EFAULT; |
3368 | if (tmp == BRCTL_GET_VERSION) | |
3369 | return BRCTL_VERSION + 1; | |
3370 | return -EINVAL; | |
3371 | } | |
3372 | ||
6b96018b AB |
3373 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3374 | unsigned int cmd, unsigned long arg) | |
3375 | { | |
3376 | void __user *argp = compat_ptr(arg); | |
3377 | struct sock *sk = sock->sk; | |
3378 | struct net *net = sock_net(sk); | |
7a229387 | 3379 | |
6b96018b | 3380 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3381 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3382 | |
3383 | switch (cmd) { | |
3384 | case SIOCSIFBR: | |
3385 | case SIOCGIFBR: | |
3386 | return old_bridge_ioctl(argp); | |
6b96018b | 3387 | case SIOCGIFCONF: |
36fd633e | 3388 | return compat_dev_ifconf(net, argp); |
6b96018b AB |
3389 | case SIOCETHTOOL: |
3390 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3391 | case SIOCWANDEV: |
3392 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3393 | case SIOCGIFMAP: |
3394 | case SIOCSIFMAP: | |
3395 | return compat_sioc_ifmap(net, cmd, argp); | |
0768e170 AB |
3396 | case SIOCGSTAMP_OLD: |
3397 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
3398 | if (!sock->ops->gettstamp) |
3399 | return -ENOIOCTLCMD; | |
0768e170 | 3400 | return sock->ops->gettstamp(sock, argp, cmd == SIOCGSTAMP_OLD, |
c7cbdbf2 AB |
3401 | !COMPAT_USE_64BIT_TIME); |
3402 | ||
590d4693 BH |
3403 | case SIOCBONDSLAVEINFOQUERY: |
3404 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3405 | case SIOCSHWTSTAMP: |
fd468c74 | 3406 | case SIOCGHWTSTAMP: |
590d4693 | 3407 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3408 | |
3409 | case FIOSETOWN: | |
3410 | case SIOCSPGRP: | |
3411 | case FIOGETOWN: | |
3412 | case SIOCGPGRP: | |
3413 | case SIOCBRADDBR: | |
3414 | case SIOCBRDELBR: | |
3415 | case SIOCGIFVLAN: | |
3416 | case SIOCSIFVLAN: | |
c62cce2c | 3417 | case SIOCGSKNS: |
0768e170 AB |
3418 | case SIOCGSTAMP_NEW: |
3419 | case SIOCGSTAMPNS_NEW: | |
6b96018b AB |
3420 | return sock_ioctl(file, cmd, arg); |
3421 | ||
3422 | case SIOCGIFFLAGS: | |
3423 | case SIOCSIFFLAGS: | |
3424 | case SIOCGIFMETRIC: | |
3425 | case SIOCSIFMETRIC: | |
3426 | case SIOCGIFMTU: | |
3427 | case SIOCSIFMTU: | |
3428 | case SIOCGIFMEM: | |
3429 | case SIOCSIFMEM: | |
3430 | case SIOCGIFHWADDR: | |
3431 | case SIOCSIFHWADDR: | |
3432 | case SIOCADDMULTI: | |
3433 | case SIOCDELMULTI: | |
3434 | case SIOCGIFINDEX: | |
6b96018b AB |
3435 | case SIOCGIFADDR: |
3436 | case SIOCSIFADDR: | |
3437 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3438 | case SIOCDIFADDR: |
6b96018b AB |
3439 | case SIOCGIFBRDADDR: |
3440 | case SIOCSIFBRDADDR: | |
3441 | case SIOCGIFDSTADDR: | |
3442 | case SIOCSIFDSTADDR: | |
3443 | case SIOCGIFNETMASK: | |
3444 | case SIOCSIFNETMASK: | |
3445 | case SIOCSIFPFLAGS: | |
3446 | case SIOCGIFPFLAGS: | |
3447 | case SIOCGIFTXQLEN: | |
3448 | case SIOCSIFTXQLEN: | |
3449 | case SIOCBRADDIF: | |
3450 | case SIOCBRDELIF: | |
c6c9fee3 | 3451 | case SIOCGIFNAME: |
9177efd3 AB |
3452 | case SIOCSIFNAME: |
3453 | case SIOCGMIIPHY: | |
3454 | case SIOCGMIIREG: | |
3455 | case SIOCSMIIREG: | |
f92d4fc9 AV |
3456 | case SIOCBONDENSLAVE: |
3457 | case SIOCBONDRELEASE: | |
3458 | case SIOCBONDSETHWADDR: | |
3459 | case SIOCBONDCHANGEACTIVE: | |
37ac39bd JB |
3460 | return compat_ifreq_ioctl(net, sock, cmd, argp); |
3461 | ||
6b96018b AB |
3462 | case SIOCSARP: |
3463 | case SIOCGARP: | |
3464 | case SIOCDARP: | |
c7dc504e | 3465 | case SIOCOUTQ: |
9d7bf41f | 3466 | case SIOCOUTQNSD: |
6b96018b | 3467 | case SIOCATMARK: |
63ff03ab | 3468 | return sock_do_ioctl(net, sock, cmd, arg); |
9177efd3 AB |
3469 | } |
3470 | ||
6b96018b AB |
3471 | return -ENOIOCTLCMD; |
3472 | } | |
7a229387 | 3473 | |
95c96174 | 3474 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3475 | unsigned long arg) |
89bbfc95 SP |
3476 | { |
3477 | struct socket *sock = file->private_data; | |
3478 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3479 | struct sock *sk; |
3480 | struct net *net; | |
3481 | ||
3482 | sk = sock->sk; | |
3483 | net = sock_net(sk); | |
89bbfc95 SP |
3484 | |
3485 | if (sock->ops->compat_ioctl) | |
3486 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3487 | ||
87de87d5 DM |
3488 | if (ret == -ENOIOCTLCMD && |
3489 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3490 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3491 | ||
6b96018b AB |
3492 | if (ret == -ENOIOCTLCMD) |
3493 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3494 | ||
89bbfc95 SP |
3495 | return ret; |
3496 | } | |
3497 | #endif | |
3498 | ||
8a3c245c PT |
3499 | /** |
3500 | * kernel_bind - bind an address to a socket (kernel space) | |
3501 | * @sock: socket | |
3502 | * @addr: address | |
3503 | * @addrlen: length of address | |
3504 | * | |
3505 | * Returns 0 or an error. | |
3506 | */ | |
3507 | ||
ac5a488e SS |
3508 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3509 | { | |
3510 | return sock->ops->bind(sock, addr, addrlen); | |
3511 | } | |
c6d409cf | 3512 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e | 3513 | |
8a3c245c PT |
3514 | /** |
3515 | * kernel_listen - move socket to listening state (kernel space) | |
3516 | * @sock: socket | |
3517 | * @backlog: pending connections queue size | |
3518 | * | |
3519 | * Returns 0 or an error. | |
3520 | */ | |
3521 | ||
ac5a488e SS |
3522 | int kernel_listen(struct socket *sock, int backlog) |
3523 | { | |
3524 | return sock->ops->listen(sock, backlog); | |
3525 | } | |
c6d409cf | 3526 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e | 3527 | |
8a3c245c PT |
3528 | /** |
3529 | * kernel_accept - accept a connection (kernel space) | |
3530 | * @sock: listening socket | |
3531 | * @newsock: new connected socket | |
3532 | * @flags: flags | |
3533 | * | |
3534 | * @flags must be SOCK_CLOEXEC, SOCK_NONBLOCK or 0. | |
3535 | * If it fails, @newsock is guaranteed to be %NULL. | |
3536 | * Returns 0 or an error. | |
3537 | */ | |
3538 | ||
ac5a488e SS |
3539 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) |
3540 | { | |
3541 | struct sock *sk = sock->sk; | |
3542 | int err; | |
3543 | ||
3544 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3545 | newsock); | |
3546 | if (err < 0) | |
3547 | goto done; | |
3548 | ||
cdfbabfb | 3549 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3550 | if (err < 0) { |
3551 | sock_release(*newsock); | |
fa8705b0 | 3552 | *newsock = NULL; |
ac5a488e SS |
3553 | goto done; |
3554 | } | |
3555 | ||
3556 | (*newsock)->ops = sock->ops; | |
1b08534e | 3557 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3558 | |
3559 | done: | |
3560 | return err; | |
3561 | } | |
c6d409cf | 3562 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e | 3563 | |
8a3c245c PT |
3564 | /** |
3565 | * kernel_connect - connect a socket (kernel space) | |
3566 | * @sock: socket | |
3567 | * @addr: address | |
3568 | * @addrlen: address length | |
3569 | * @flags: flags (O_NONBLOCK, ...) | |
3570 | * | |
f1dcffcc | 3571 | * For datagram sockets, @addr is the address to which datagrams are sent |
8a3c245c PT |
3572 | * by default, and the only address from which datagrams are received. |
3573 | * For stream sockets, attempts to connect to @addr. | |
3574 | * Returns 0 or an error code. | |
3575 | */ | |
3576 | ||
ac5a488e | 3577 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, |
4768fbcb | 3578 | int flags) |
ac5a488e SS |
3579 | { |
3580 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3581 | } | |
c6d409cf | 3582 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e | 3583 | |
8a3c245c PT |
3584 | /** |
3585 | * kernel_getsockname - get the address which the socket is bound (kernel space) | |
3586 | * @sock: socket | |
3587 | * @addr: address holder | |
3588 | * | |
3589 | * Fills the @addr pointer with the address which the socket is bound. | |
3590 | * Returns 0 or an error code. | |
3591 | */ | |
3592 | ||
9b2c45d4 | 3593 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3594 | { |
9b2c45d4 | 3595 | return sock->ops->getname(sock, addr, 0); |
ac5a488e | 3596 | } |
c6d409cf | 3597 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e | 3598 | |
8a3c245c | 3599 | /** |
645f0897 | 3600 | * kernel_getpeername - get the address which the socket is connected (kernel space) |
8a3c245c PT |
3601 | * @sock: socket |
3602 | * @addr: address holder | |
3603 | * | |
3604 | * Fills the @addr pointer with the address which the socket is connected. | |
3605 | * Returns 0 or an error code. | |
3606 | */ | |
3607 | ||
9b2c45d4 | 3608 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3609 | { |
9b2c45d4 | 3610 | return sock->ops->getname(sock, addr, 1); |
ac5a488e | 3611 | } |
c6d409cf | 3612 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e | 3613 | |
8a3c245c PT |
3614 | /** |
3615 | * kernel_sendpage - send a &page through a socket (kernel space) | |
3616 | * @sock: socket | |
3617 | * @page: page | |
3618 | * @offset: page offset | |
3619 | * @size: total size in bytes | |
3620 | * @flags: flags (MSG_DONTWAIT, ...) | |
3621 | * | |
3622 | * Returns the total amount sent in bytes or an error. | |
3623 | */ | |
3624 | ||
ac5a488e SS |
3625 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, |
3626 | size_t size, int flags) | |
3627 | { | |
7b62d31d CL |
3628 | if (sock->ops->sendpage) { |
3629 | /* Warn in case the improper page to zero-copy send */ | |
3630 | WARN_ONCE(!sendpage_ok(page), "improper page for zero-copy send"); | |
ac5a488e | 3631 | return sock->ops->sendpage(sock, page, offset, size, flags); |
7b62d31d | 3632 | } |
ac5a488e SS |
3633 | return sock_no_sendpage(sock, page, offset, size, flags); |
3634 | } | |
c6d409cf | 3635 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3636 | |
8a3c245c PT |
3637 | /** |
3638 | * kernel_sendpage_locked - send a &page through the locked sock (kernel space) | |
3639 | * @sk: sock | |
3640 | * @page: page | |
3641 | * @offset: page offset | |
3642 | * @size: total size in bytes | |
3643 | * @flags: flags (MSG_DONTWAIT, ...) | |
3644 | * | |
3645 | * Returns the total amount sent in bytes or an error. | |
3646 | * Caller must hold @sk. | |
3647 | */ | |
3648 | ||
306b13eb TH |
3649 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3650 | size_t size, int flags) | |
3651 | { | |
3652 | struct socket *sock = sk->sk_socket; | |
3653 | ||
3654 | if (sock->ops->sendpage_locked) | |
3655 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3656 | flags); | |
3657 | ||
3658 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3659 | } | |
3660 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3661 | ||
8a3c245c | 3662 | /** |
645f0897 | 3663 | * kernel_sock_shutdown - shut down part of a full-duplex connection (kernel space) |
8a3c245c PT |
3664 | * @sock: socket |
3665 | * @how: connection part | |
3666 | * | |
3667 | * Returns 0 or an error. | |
3668 | */ | |
3669 | ||
91cf45f0 TM |
3670 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3671 | { | |
3672 | return sock->ops->shutdown(sock, how); | |
3673 | } | |
91cf45f0 | 3674 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 | 3675 | |
8a3c245c PT |
3676 | /** |
3677 | * kernel_sock_ip_overhead - returns the IP overhead imposed by a socket | |
3678 | * @sk: socket | |
3679 | * | |
3680 | * This routine returns the IP overhead imposed by a socket i.e. | |
3681 | * the length of the underlying IP header, depending on whether | |
3682 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
3683 | * on at the socket. Assumes that the caller has a lock on the socket. | |
113c3075 | 3684 | */ |
8a3c245c | 3685 | |
113c3075 P |
3686 | u32 kernel_sock_ip_overhead(struct sock *sk) |
3687 | { | |
3688 | struct inet_sock *inet; | |
3689 | struct ip_options_rcu *opt; | |
3690 | u32 overhead = 0; | |
113c3075 P |
3691 | #if IS_ENABLED(CONFIG_IPV6) |
3692 | struct ipv6_pinfo *np; | |
3693 | struct ipv6_txoptions *optv6 = NULL; | |
3694 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3695 | ||
3696 | if (!sk) | |
3697 | return overhead; | |
3698 | ||
113c3075 P |
3699 | switch (sk->sk_family) { |
3700 | case AF_INET: | |
3701 | inet = inet_sk(sk); | |
3702 | overhead += sizeof(struct iphdr); | |
3703 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3704 | sock_owned_by_user(sk)); |
113c3075 P |
3705 | if (opt) |
3706 | overhead += opt->opt.optlen; | |
3707 | return overhead; | |
3708 | #if IS_ENABLED(CONFIG_IPV6) | |
3709 | case AF_INET6: | |
3710 | np = inet6_sk(sk); | |
3711 | overhead += sizeof(struct ipv6hdr); | |
3712 | if (np) | |
3713 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3714 | sock_owned_by_user(sk)); |
113c3075 P |
3715 | if (optv6) |
3716 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3717 | return overhead; | |
3718 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3719 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3720 | return overhead; | |
3721 | } | |
3722 | } | |
3723 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |