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> |
d7c08826 | 107 | #include <linux/ptp_clock_kernel.h> |
06021292 | 108 | |
e0d1095a | 109 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
110 | unsigned int sysctl_net_busy_read __read_mostly; |
111 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 112 | #endif |
6b96018b | 113 | |
8ae5e030 AV |
114 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
115 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 116 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
117 | |
118 | static int sock_close(struct inode *inode, struct file *file); | |
a11e1d43 LT |
119 | static __poll_t sock_poll(struct file *file, |
120 | struct poll_table_struct *wait); | |
89bddce5 | 121 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
122 | #ifdef CONFIG_COMPAT |
123 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 124 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 125 | #endif |
1da177e4 | 126 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
127 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
128 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 129 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 130 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 131 | unsigned int flags); |
542d3065 AB |
132 | |
133 | #ifdef CONFIG_PROC_FS | |
134 | static void sock_show_fdinfo(struct seq_file *m, struct file *f) | |
135 | { | |
136 | struct socket *sock = f->private_data; | |
137 | ||
138 | if (sock->ops->show_fdinfo) | |
139 | sock->ops->show_fdinfo(m, sock); | |
140 | } | |
141 | #else | |
142 | #define sock_show_fdinfo NULL | |
143 | #endif | |
1da177e4 | 144 | |
1da177e4 LT |
145 | /* |
146 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
147 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
148 | */ | |
149 | ||
da7071d7 | 150 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
151 | .owner = THIS_MODULE, |
152 | .llseek = no_llseek, | |
8ae5e030 AV |
153 | .read_iter = sock_read_iter, |
154 | .write_iter = sock_write_iter, | |
1da177e4 LT |
155 | .poll = sock_poll, |
156 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
157 | #ifdef CONFIG_COMPAT |
158 | .compat_ioctl = compat_sock_ioctl, | |
159 | #endif | |
1da177e4 | 160 | .mmap = sock_mmap, |
1da177e4 LT |
161 | .release = sock_close, |
162 | .fasync = sock_fasync, | |
5274f052 JA |
163 | .sendpage = sock_sendpage, |
164 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 165 | .splice_read = sock_splice_read, |
b4653342 | 166 | .show_fdinfo = sock_show_fdinfo, |
1da177e4 LT |
167 | }; |
168 | ||
fe0bdbde YD |
169 | static const char * const pf_family_names[] = { |
170 | [PF_UNSPEC] = "PF_UNSPEC", | |
171 | [PF_UNIX] = "PF_UNIX/PF_LOCAL", | |
172 | [PF_INET] = "PF_INET", | |
173 | [PF_AX25] = "PF_AX25", | |
174 | [PF_IPX] = "PF_IPX", | |
175 | [PF_APPLETALK] = "PF_APPLETALK", | |
176 | [PF_NETROM] = "PF_NETROM", | |
177 | [PF_BRIDGE] = "PF_BRIDGE", | |
178 | [PF_ATMPVC] = "PF_ATMPVC", | |
179 | [PF_X25] = "PF_X25", | |
180 | [PF_INET6] = "PF_INET6", | |
181 | [PF_ROSE] = "PF_ROSE", | |
182 | [PF_DECnet] = "PF_DECnet", | |
183 | [PF_NETBEUI] = "PF_NETBEUI", | |
184 | [PF_SECURITY] = "PF_SECURITY", | |
185 | [PF_KEY] = "PF_KEY", | |
186 | [PF_NETLINK] = "PF_NETLINK/PF_ROUTE", | |
187 | [PF_PACKET] = "PF_PACKET", | |
188 | [PF_ASH] = "PF_ASH", | |
189 | [PF_ECONET] = "PF_ECONET", | |
190 | [PF_ATMSVC] = "PF_ATMSVC", | |
191 | [PF_RDS] = "PF_RDS", | |
192 | [PF_SNA] = "PF_SNA", | |
193 | [PF_IRDA] = "PF_IRDA", | |
194 | [PF_PPPOX] = "PF_PPPOX", | |
195 | [PF_WANPIPE] = "PF_WANPIPE", | |
196 | [PF_LLC] = "PF_LLC", | |
197 | [PF_IB] = "PF_IB", | |
198 | [PF_MPLS] = "PF_MPLS", | |
199 | [PF_CAN] = "PF_CAN", | |
200 | [PF_TIPC] = "PF_TIPC", | |
201 | [PF_BLUETOOTH] = "PF_BLUETOOTH", | |
202 | [PF_IUCV] = "PF_IUCV", | |
203 | [PF_RXRPC] = "PF_RXRPC", | |
204 | [PF_ISDN] = "PF_ISDN", | |
205 | [PF_PHONET] = "PF_PHONET", | |
206 | [PF_IEEE802154] = "PF_IEEE802154", | |
207 | [PF_CAIF] = "PF_CAIF", | |
208 | [PF_ALG] = "PF_ALG", | |
209 | [PF_NFC] = "PF_NFC", | |
210 | [PF_VSOCK] = "PF_VSOCK", | |
211 | [PF_KCM] = "PF_KCM", | |
212 | [PF_QIPCRTR] = "PF_QIPCRTR", | |
213 | [PF_SMC] = "PF_SMC", | |
214 | [PF_XDP] = "PF_XDP", | |
215 | }; | |
216 | ||
1da177e4 LT |
217 | /* |
218 | * The protocol list. Each protocol is registered in here. | |
219 | */ | |
220 | ||
1da177e4 | 221 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 222 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 223 | |
1da177e4 | 224 | /* |
89bddce5 SH |
225 | * Support routines. |
226 | * Move socket addresses back and forth across the kernel/user | |
227 | * divide and look after the messy bits. | |
1da177e4 LT |
228 | */ |
229 | ||
1da177e4 LT |
230 | /** |
231 | * move_addr_to_kernel - copy a socket address into kernel space | |
232 | * @uaddr: Address in user space | |
233 | * @kaddr: Address in kernel space | |
234 | * @ulen: Length in user space | |
235 | * | |
236 | * The address is copied into kernel space. If the provided address is | |
237 | * too long an error code of -EINVAL is returned. If the copy gives | |
238 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
239 | */ | |
240 | ||
43db362d | 241 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 242 | { |
230b1839 | 243 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 244 | return -EINVAL; |
89bddce5 | 245 | if (ulen == 0) |
1da177e4 | 246 | return 0; |
89bddce5 | 247 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 248 | return -EFAULT; |
3ec3b2fb | 249 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
250 | } |
251 | ||
252 | /** | |
253 | * move_addr_to_user - copy an address to user space | |
254 | * @kaddr: kernel space address | |
255 | * @klen: length of address in kernel | |
256 | * @uaddr: user space address | |
257 | * @ulen: pointer to user length field | |
258 | * | |
259 | * The value pointed to by ulen on entry is the buffer length available. | |
260 | * This is overwritten with the buffer space used. -EINVAL is returned | |
261 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
262 | * is returned if either the buffer or the length field are not | |
263 | * accessible. | |
264 | * After copying the data up to the limit the user specifies, the true | |
265 | * length of the data is written over the length limit the user | |
266 | * specified. Zero is returned for a success. | |
267 | */ | |
89bddce5 | 268 | |
43db362d | 269 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 270 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
271 | { |
272 | int err; | |
273 | int len; | |
274 | ||
68c6beb3 | 275 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
276 | err = get_user(len, ulen); |
277 | if (err) | |
1da177e4 | 278 | return err; |
89bddce5 SH |
279 | if (len > klen) |
280 | len = klen; | |
68c6beb3 | 281 | if (len < 0) |
1da177e4 | 282 | return -EINVAL; |
89bddce5 | 283 | if (len) { |
d6fe3945 SG |
284 | if (audit_sockaddr(klen, kaddr)) |
285 | return -ENOMEM; | |
89bddce5 | 286 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
287 | return -EFAULT; |
288 | } | |
289 | /* | |
89bddce5 SH |
290 | * "fromlen shall refer to the value before truncation.." |
291 | * 1003.1g | |
1da177e4 LT |
292 | */ |
293 | return __put_user(klen, ulen); | |
294 | } | |
295 | ||
08009a76 | 296 | static struct kmem_cache *sock_inode_cachep __ro_after_init; |
1da177e4 LT |
297 | |
298 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
299 | { | |
300 | struct socket_alloc *ei; | |
89bddce5 | 301 | |
e94b1766 | 302 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
303 | if (!ei) |
304 | return NULL; | |
333f7909 AV |
305 | init_waitqueue_head(&ei->socket.wq.wait); |
306 | ei->socket.wq.fasync_list = NULL; | |
307 | ei->socket.wq.flags = 0; | |
89bddce5 | 308 | |
1da177e4 LT |
309 | ei->socket.state = SS_UNCONNECTED; |
310 | ei->socket.flags = 0; | |
311 | ei->socket.ops = NULL; | |
312 | ei->socket.sk = NULL; | |
313 | ei->socket.file = NULL; | |
1da177e4 LT |
314 | |
315 | return &ei->vfs_inode; | |
316 | } | |
317 | ||
6d7855c5 | 318 | static void sock_free_inode(struct inode *inode) |
1da177e4 | 319 | { |
43815482 ED |
320 | struct socket_alloc *ei; |
321 | ||
322 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
43815482 | 323 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
324 | } |
325 | ||
51cc5068 | 326 | static void init_once(void *foo) |
1da177e4 | 327 | { |
89bddce5 | 328 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 329 | |
a35afb83 | 330 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 331 | } |
89bddce5 | 332 | |
1e911632 | 333 | static void init_inodecache(void) |
1da177e4 LT |
334 | { |
335 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
336 | sizeof(struct socket_alloc), |
337 | 0, | |
338 | (SLAB_HWCACHE_ALIGN | | |
339 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 340 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 341 | init_once); |
1e911632 | 342 | BUG_ON(sock_inode_cachep == NULL); |
1da177e4 LT |
343 | } |
344 | ||
b87221de | 345 | static const struct super_operations sockfs_ops = { |
c6d409cf | 346 | .alloc_inode = sock_alloc_inode, |
6d7855c5 | 347 | .free_inode = sock_free_inode, |
c6d409cf | 348 | .statfs = simple_statfs, |
1da177e4 LT |
349 | }; |
350 | ||
c23fbb6b ED |
351 | /* |
352 | * sockfs_dname() is called from d_path(). | |
353 | */ | |
354 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
355 | { | |
356 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 357 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
358 | } |
359 | ||
3ba13d17 | 360 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 361 | .d_dname = sockfs_dname, |
1da177e4 LT |
362 | }; |
363 | ||
bba0bd31 AG |
364 | static int sockfs_xattr_get(const struct xattr_handler *handler, |
365 | struct dentry *dentry, struct inode *inode, | |
366 | const char *suffix, void *value, size_t size) | |
367 | { | |
368 | if (value) { | |
369 | if (dentry->d_name.len + 1 > size) | |
370 | return -ERANGE; | |
371 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
372 | } | |
373 | return dentry->d_name.len + 1; | |
374 | } | |
375 | ||
376 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" | |
377 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
378 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
379 | ||
380 | static const struct xattr_handler sockfs_xattr_handler = { | |
381 | .name = XATTR_NAME_SOCKPROTONAME, | |
382 | .get = sockfs_xattr_get, | |
383 | }; | |
384 | ||
4a590153 | 385 | static int sockfs_security_xattr_set(const struct xattr_handler *handler, |
e65ce2a5 | 386 | struct user_namespace *mnt_userns, |
4a590153 AG |
387 | struct dentry *dentry, struct inode *inode, |
388 | const char *suffix, const void *value, | |
389 | size_t size, int flags) | |
390 | { | |
391 | /* Handled by LSM. */ | |
392 | return -EAGAIN; | |
393 | } | |
394 | ||
395 | static const struct xattr_handler sockfs_security_xattr_handler = { | |
396 | .prefix = XATTR_SECURITY_PREFIX, | |
397 | .set = sockfs_security_xattr_set, | |
398 | }; | |
399 | ||
bba0bd31 AG |
400 | static const struct xattr_handler *sockfs_xattr_handlers[] = { |
401 | &sockfs_xattr_handler, | |
4a590153 | 402 | &sockfs_security_xattr_handler, |
bba0bd31 AG |
403 | NULL |
404 | }; | |
405 | ||
fba9be49 | 406 | static int sockfs_init_fs_context(struct fs_context *fc) |
c74a1cbb | 407 | { |
fba9be49 DH |
408 | struct pseudo_fs_context *ctx = init_pseudo(fc, SOCKFS_MAGIC); |
409 | if (!ctx) | |
410 | return -ENOMEM; | |
411 | ctx->ops = &sockfs_ops; | |
412 | ctx->dops = &sockfs_dentry_operations; | |
413 | ctx->xattr = sockfs_xattr_handlers; | |
414 | return 0; | |
c74a1cbb AV |
415 | } |
416 | ||
417 | static struct vfsmount *sock_mnt __read_mostly; | |
418 | ||
419 | static struct file_system_type sock_fs_type = { | |
420 | .name = "sockfs", | |
fba9be49 | 421 | .init_fs_context = sockfs_init_fs_context, |
c74a1cbb AV |
422 | .kill_sb = kill_anon_super, |
423 | }; | |
424 | ||
1da177e4 LT |
425 | /* |
426 | * Obtains the first available file descriptor and sets it up for use. | |
427 | * | |
39d8c1b6 DM |
428 | * These functions create file structures and maps them to fd space |
429 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
430 | * and file struct implicitly stored in sock->file. |
431 | * Note that another thread may close file descriptor before we return | |
432 | * from this function. We use the fact that now we do not refer | |
433 | * to socket after mapping. If one day we will need it, this | |
434 | * function will increment ref. count on file by 1. | |
435 | * | |
436 | * In any case returned fd MAY BE not valid! | |
437 | * This race condition is unavoidable | |
438 | * with shared fd spaces, we cannot solve it inside kernel, | |
439 | * but we take care of internal coherence yet. | |
440 | */ | |
441 | ||
8a3c245c PT |
442 | /** |
443 | * sock_alloc_file - Bind a &socket to a &file | |
444 | * @sock: socket | |
445 | * @flags: file status flags | |
446 | * @dname: protocol name | |
447 | * | |
448 | * Returns the &file bound with @sock, implicitly storing it | |
449 | * in sock->file. If dname is %NULL, sets to "". | |
450 | * On failure the return is a ERR pointer (see linux/err.h). | |
451 | * This function uses GFP_KERNEL internally. | |
452 | */ | |
453 | ||
aab174f0 | 454 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 455 | { |
7cbe66b6 | 456 | struct file *file; |
1da177e4 | 457 | |
d93aa9d8 AV |
458 | if (!dname) |
459 | dname = sock->sk ? sock->sk->sk_prot_creator->name : ""; | |
39d8c1b6 | 460 | |
d93aa9d8 AV |
461 | file = alloc_file_pseudo(SOCK_INODE(sock), sock_mnt, dname, |
462 | O_RDWR | (flags & O_NONBLOCK), | |
463 | &socket_file_ops); | |
b5ffe634 | 464 | if (IS_ERR(file)) { |
8e1611e2 | 465 | sock_release(sock); |
39b65252 | 466 | return file; |
cc3808f8 AV |
467 | } |
468 | ||
469 | sock->file = file; | |
39d8c1b6 | 470 | file->private_data = sock; |
d8e464ec | 471 | stream_open(SOCK_INODE(sock), file); |
28407630 | 472 | return file; |
39d8c1b6 | 473 | } |
56b31d1c | 474 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 475 | |
56b31d1c | 476 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
477 | { |
478 | struct file *newfile; | |
28407630 | 479 | int fd = get_unused_fd_flags(flags); |
ce4bb04c AV |
480 | if (unlikely(fd < 0)) { |
481 | sock_release(sock); | |
28407630 | 482 | return fd; |
ce4bb04c | 483 | } |
39d8c1b6 | 484 | |
aab174f0 | 485 | newfile = sock_alloc_file(sock, flags, NULL); |
4546e44c | 486 | if (!IS_ERR(newfile)) { |
39d8c1b6 | 487 | fd_install(fd, newfile); |
28407630 AV |
488 | return fd; |
489 | } | |
7cbe66b6 | 490 | |
28407630 AV |
491 | put_unused_fd(fd); |
492 | return PTR_ERR(newfile); | |
1da177e4 LT |
493 | } |
494 | ||
8a3c245c PT |
495 | /** |
496 | * sock_from_file - Return the &socket bounded to @file. | |
497 | * @file: file | |
8a3c245c | 498 | * |
dba4a925 | 499 | * On failure returns %NULL. |
8a3c245c PT |
500 | */ |
501 | ||
dba4a925 | 502 | struct socket *sock_from_file(struct file *file) |
6cb153ca | 503 | { |
6cb153ca BL |
504 | if (file->f_op == &socket_file_ops) |
505 | return file->private_data; /* set in sock_map_fd */ | |
506 | ||
23bb80d2 | 507 | return NULL; |
6cb153ca | 508 | } |
406a3c63 | 509 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 510 | |
1da177e4 | 511 | /** |
c6d409cf | 512 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
513 | * @fd: file handle |
514 | * @err: pointer to an error code return | |
515 | * | |
516 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 517 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
518 | * with a negative errno code and NULL is returned. The function checks |
519 | * for both invalid handles and passing a handle which is not a socket. | |
520 | * | |
521 | * On a success the socket object pointer is returned. | |
522 | */ | |
523 | ||
524 | struct socket *sockfd_lookup(int fd, int *err) | |
525 | { | |
526 | struct file *file; | |
1da177e4 LT |
527 | struct socket *sock; |
528 | ||
89bddce5 SH |
529 | file = fget(fd); |
530 | if (!file) { | |
1da177e4 LT |
531 | *err = -EBADF; |
532 | return NULL; | |
533 | } | |
89bddce5 | 534 | |
dba4a925 FR |
535 | sock = sock_from_file(file); |
536 | if (!sock) { | |
537 | *err = -ENOTSOCK; | |
1da177e4 | 538 | fput(file); |
dba4a925 | 539 | } |
6cb153ca BL |
540 | return sock; |
541 | } | |
c6d409cf | 542 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 543 | |
6cb153ca BL |
544 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
545 | { | |
00e188ef | 546 | struct fd f = fdget(fd); |
6cb153ca BL |
547 | struct socket *sock; |
548 | ||
3672558c | 549 | *err = -EBADF; |
00e188ef | 550 | if (f.file) { |
dba4a925 | 551 | sock = sock_from_file(f.file); |
00e188ef | 552 | if (likely(sock)) { |
ce787a5a | 553 | *fput_needed = f.flags & FDPUT_FPUT; |
6cb153ca | 554 | return sock; |
00e188ef | 555 | } |
dba4a925 | 556 | *err = -ENOTSOCK; |
00e188ef | 557 | fdput(f); |
1da177e4 | 558 | } |
6cb153ca | 559 | return NULL; |
1da177e4 LT |
560 | } |
561 | ||
600e1779 MY |
562 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
563 | size_t size) | |
564 | { | |
565 | ssize_t len; | |
566 | ssize_t used = 0; | |
567 | ||
c5ef6035 | 568 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
569 | if (len < 0) |
570 | return len; | |
571 | used += len; | |
572 | if (buffer) { | |
573 | if (size < used) | |
574 | return -ERANGE; | |
575 | buffer += len; | |
576 | } | |
577 | ||
578 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
579 | used += len; | |
580 | if (buffer) { | |
581 | if (size < used) | |
582 | return -ERANGE; | |
583 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
584 | buffer += len; | |
585 | } | |
586 | ||
587 | return used; | |
588 | } | |
589 | ||
549c7297 CB |
590 | static int sockfs_setattr(struct user_namespace *mnt_userns, |
591 | struct dentry *dentry, struct iattr *iattr) | |
86741ec2 | 592 | { |
549c7297 | 593 | int err = simple_setattr(&init_user_ns, dentry, iattr); |
86741ec2 | 594 | |
e1a3a60a | 595 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
596 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
597 | ||
6d8c50dc CW |
598 | if (sock->sk) |
599 | sock->sk->sk_uid = iattr->ia_uid; | |
600 | else | |
601 | err = -ENOENT; | |
86741ec2 LC |
602 | } |
603 | ||
604 | return err; | |
605 | } | |
606 | ||
600e1779 | 607 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 608 | .listxattr = sockfs_listxattr, |
86741ec2 | 609 | .setattr = sockfs_setattr, |
600e1779 MY |
610 | }; |
611 | ||
1da177e4 | 612 | /** |
8a3c245c | 613 | * sock_alloc - allocate a socket |
89bddce5 | 614 | * |
1da177e4 LT |
615 | * Allocate a new inode and socket object. The two are bound together |
616 | * and initialised. The socket is then returned. If we are out of inodes | |
8a3c245c | 617 | * NULL is returned. This functions uses GFP_KERNEL internally. |
1da177e4 LT |
618 | */ |
619 | ||
f4a00aac | 620 | struct socket *sock_alloc(void) |
1da177e4 | 621 | { |
89bddce5 SH |
622 | struct inode *inode; |
623 | struct socket *sock; | |
1da177e4 | 624 | |
a209dfc7 | 625 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
626 | if (!inode) |
627 | return NULL; | |
628 | ||
629 | sock = SOCKET_I(inode); | |
630 | ||
85fe4025 | 631 | inode->i_ino = get_next_ino(); |
89bddce5 | 632 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
633 | inode->i_uid = current_fsuid(); |
634 | inode->i_gid = current_fsgid(); | |
600e1779 | 635 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 636 | |
1da177e4 LT |
637 | return sock; |
638 | } | |
f4a00aac | 639 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 640 | |
6d8c50dc | 641 | static void __sock_release(struct socket *sock, struct inode *inode) |
1da177e4 LT |
642 | { |
643 | if (sock->ops) { | |
644 | struct module *owner = sock->ops->owner; | |
645 | ||
6d8c50dc CW |
646 | if (inode) |
647 | inode_lock(inode); | |
1da177e4 | 648 | sock->ops->release(sock); |
ff7b11aa | 649 | sock->sk = NULL; |
6d8c50dc CW |
650 | if (inode) |
651 | inode_unlock(inode); | |
1da177e4 LT |
652 | sock->ops = NULL; |
653 | module_put(owner); | |
654 | } | |
655 | ||
333f7909 | 656 | if (sock->wq.fasync_list) |
3410f22e | 657 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 658 | |
1da177e4 LT |
659 | if (!sock->file) { |
660 | iput(SOCK_INODE(sock)); | |
661 | return; | |
662 | } | |
89bddce5 | 663 | sock->file = NULL; |
1da177e4 | 664 | } |
6d8c50dc | 665 | |
9a8ad9ac AL |
666 | /** |
667 | * sock_release - close a socket | |
668 | * @sock: socket to close | |
669 | * | |
670 | * The socket is released from the protocol stack if it has a release | |
671 | * callback, and the inode is then released if the socket is bound to | |
672 | * an inode not a file. | |
673 | */ | |
6d8c50dc CW |
674 | void sock_release(struct socket *sock) |
675 | { | |
676 | __sock_release(sock, NULL); | |
677 | } | |
c6d409cf | 678 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 679 | |
c14ac945 | 680 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 681 | { |
140c55d4 ED |
682 | u8 flags = *tx_flags; |
683 | ||
c14ac945 | 684 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
685 | flags |= SKBTX_HW_TSTAMP; |
686 | ||
c14ac945 | 687 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
688 | flags |= SKBTX_SW_TSTAMP; |
689 | ||
c14ac945 | 690 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
691 | flags |= SKBTX_SCHED_TSTAMP; |
692 | ||
140c55d4 | 693 | *tx_flags = flags; |
20d49473 | 694 | } |
67cc0d40 | 695 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 696 | |
8c3c447b PA |
697 | INDIRECT_CALLABLE_DECLARE(int inet_sendmsg(struct socket *, struct msghdr *, |
698 | size_t)); | |
a648a592 PA |
699 | INDIRECT_CALLABLE_DECLARE(int inet6_sendmsg(struct socket *, struct msghdr *, |
700 | size_t)); | |
d8725c86 | 701 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 702 | { |
a648a592 PA |
703 | int ret = INDIRECT_CALL_INET(sock->ops->sendmsg, inet6_sendmsg, |
704 | inet_sendmsg, sock, msg, | |
705 | msg_data_left(msg)); | |
d8725c86 AV |
706 | BUG_ON(ret == -EIOCBQUEUED); |
707 | return ret; | |
1da177e4 LT |
708 | } |
709 | ||
85806af0 RD |
710 | /** |
711 | * sock_sendmsg - send a message through @sock | |
712 | * @sock: socket | |
713 | * @msg: message to send | |
714 | * | |
715 | * Sends @msg through @sock, passing through LSM. | |
716 | * Returns the number of bytes sent, or an error code. | |
717 | */ | |
d8725c86 | 718 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 719 | { |
d8725c86 | 720 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 721 | msg_data_left(msg)); |
228e548e | 722 | |
d8725c86 | 723 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 724 | } |
c6d409cf | 725 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 | 726 | |
8a3c245c PT |
727 | /** |
728 | * kernel_sendmsg - send a message through @sock (kernel-space) | |
729 | * @sock: socket | |
730 | * @msg: message header | |
731 | * @vec: kernel vec | |
732 | * @num: vec array length | |
733 | * @size: total message data size | |
734 | * | |
735 | * Builds the message data with @vec and sends it through @sock. | |
736 | * Returns the number of bytes sent, or an error code. | |
737 | */ | |
738 | ||
1da177e4 LT |
739 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
740 | struct kvec *vec, size_t num, size_t size) | |
741 | { | |
aa563d7b | 742 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
d8725c86 | 743 | return sock_sendmsg(sock, msg); |
1da177e4 | 744 | } |
c6d409cf | 745 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 746 | |
8a3c245c PT |
747 | /** |
748 | * kernel_sendmsg_locked - send a message through @sock (kernel-space) | |
749 | * @sk: sock | |
750 | * @msg: message header | |
751 | * @vec: output s/g array | |
752 | * @num: output s/g array length | |
753 | * @size: total message data size | |
754 | * | |
755 | * Builds the message data with @vec and sends it through @sock. | |
756 | * Returns the number of bytes sent, or an error code. | |
757 | * Caller must hold @sk. | |
758 | */ | |
759 | ||
306b13eb TH |
760 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
761 | struct kvec *vec, size_t num, size_t size) | |
762 | { | |
763 | struct socket *sock = sk->sk_socket; | |
764 | ||
765 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 766 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb | 767 | |
aa563d7b | 768 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
306b13eb TH |
769 | |
770 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
771 | } | |
772 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
773 | ||
8605330a SHY |
774 | static bool skb_is_err_queue(const struct sk_buff *skb) |
775 | { | |
776 | /* pkt_type of skbs enqueued on the error queue are set to | |
777 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
778 | * in recvmsg, since skbs received on a local socket will never | |
779 | * have a pkt_type of PACKET_OUTGOING. | |
780 | */ | |
781 | return skb->pkt_type == PACKET_OUTGOING; | |
782 | } | |
783 | ||
b50a5c70 ML |
784 | /* On transmit, software and hardware timestamps are returned independently. |
785 | * As the two skb clones share the hardware timestamp, which may be updated | |
786 | * before the software timestamp is received, a hardware TX timestamp may be | |
787 | * returned only if there is no software TX timestamp. Ignore false software | |
788 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
7f1bc6e9 | 789 | * option SO_TIMESTAMP_OLD(NS) is enabled on the socket, even when the skb has a |
b50a5c70 ML |
790 | * hardware timestamp. |
791 | */ | |
792 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
793 | { | |
794 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
795 | } | |
796 | ||
aad9c8c4 ML |
797 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb) |
798 | { | |
799 | struct scm_ts_pktinfo ts_pktinfo; | |
800 | struct net_device *orig_dev; | |
801 | ||
802 | if (!skb_mac_header_was_set(skb)) | |
803 | return; | |
804 | ||
805 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
806 | ||
807 | rcu_read_lock(); | |
808 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
809 | if (orig_dev) | |
810 | ts_pktinfo.if_index = orig_dev->ifindex; | |
811 | rcu_read_unlock(); | |
812 | ||
813 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
814 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
815 | sizeof(ts_pktinfo), &ts_pktinfo); | |
816 | } | |
817 | ||
92f37fd2 ED |
818 | /* |
819 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
820 | */ | |
821 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
822 | struct sk_buff *skb) | |
823 | { | |
20d49473 | 824 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
887feae3 | 825 | int new_tstamp = sock_flag(sk, SOCK_TSTAMP_NEW); |
9718475e DD |
826 | struct scm_timestamping_internal tss; |
827 | ||
b50a5c70 | 828 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
829 | struct skb_shared_hwtstamps *shhwtstamps = |
830 | skb_hwtstamps(skb); | |
831 | ||
832 | /* Race occurred between timestamp enabling and packet | |
833 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 834 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 835 | __net_timestamp(skb); |
b50a5c70 ML |
836 | false_tstamp = 1; |
837 | } | |
20d49473 PO |
838 | |
839 | if (need_software_tstamp) { | |
840 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
887feae3 DD |
841 | if (new_tstamp) { |
842 | struct __kernel_sock_timeval tv; | |
843 | ||
844 | skb_get_new_timestamp(skb, &tv); | |
845 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_NEW, | |
846 | sizeof(tv), &tv); | |
847 | } else { | |
848 | struct __kernel_old_timeval tv; | |
849 | ||
850 | skb_get_timestamp(skb, &tv); | |
851 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_OLD, | |
852 | sizeof(tv), &tv); | |
853 | } | |
20d49473 | 854 | } else { |
887feae3 DD |
855 | if (new_tstamp) { |
856 | struct __kernel_timespec ts; | |
857 | ||
858 | skb_get_new_timestampns(skb, &ts); | |
859 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_NEW, | |
860 | sizeof(ts), &ts); | |
861 | } else { | |
df1b4ba9 | 862 | struct __kernel_old_timespec ts; |
887feae3 DD |
863 | |
864 | skb_get_timestampns(skb, &ts); | |
865 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_OLD, | |
866 | sizeof(ts), &ts); | |
867 | } | |
20d49473 PO |
868 | } |
869 | } | |
870 | ||
f24b9be5 | 871 | memset(&tss, 0, sizeof(tss)); |
c199105d | 872 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
9718475e | 873 | ktime_to_timespec64_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 874 | empty = 0; |
4d276eb6 | 875 | if (shhwtstamps && |
b9f40e21 | 876 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
d7c08826 YL |
877 | !skb_is_swtx_tstamp(skb, false_tstamp)) { |
878 | if (sk->sk_tsflags & SOF_TIMESTAMPING_BIND_PHC) | |
879 | ptp_convert_timestamp(shhwtstamps, sk->sk_bind_phc); | |
880 | ||
881 | if (ktime_to_timespec64_cond(shhwtstamps->hwtstamp, | |
882 | tss.ts + 2)) { | |
883 | empty = 0; | |
884 | ||
885 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && | |
886 | !skb_is_err_queue(skb)) | |
887 | put_ts_pktinfo(msg, skb); | |
888 | } | |
aad9c8c4 | 889 | } |
1c885808 | 890 | if (!empty) { |
9718475e DD |
891 | if (sock_flag(sk, SOCK_TSTAMP_NEW)) |
892 | put_cmsg_scm_timestamping64(msg, &tss); | |
893 | else | |
894 | put_cmsg_scm_timestamping(msg, &tss); | |
1c885808 | 895 | |
8605330a | 896 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 897 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
898 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
899 | skb->len, skb->data); | |
900 | } | |
92f37fd2 | 901 | } |
7c81fd8b ACM |
902 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
903 | ||
6e3e939f JB |
904 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
905 | struct sk_buff *skb) | |
906 | { | |
907 | int ack; | |
908 | ||
909 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
910 | return; | |
911 | if (!skb->wifi_acked_valid) | |
912 | return; | |
913 | ||
914 | ack = skb->wifi_acked; | |
915 | ||
916 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
917 | } | |
918 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
919 | ||
11165f14 | 920 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
921 | struct sk_buff *skb) | |
3b885787 | 922 | { |
744d5a3e | 923 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 924 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 925 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
926 | } |
927 | ||
767dd033 | 928 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
929 | struct sk_buff *skb) |
930 | { | |
931 | sock_recv_timestamp(msg, sk, skb); | |
932 | sock_recv_drops(msg, sk, skb); | |
933 | } | |
767dd033 | 934 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 935 | |
8c3c447b | 936 | INDIRECT_CALLABLE_DECLARE(int inet_recvmsg(struct socket *, struct msghdr *, |
a648a592 PA |
937 | size_t, int)); |
938 | INDIRECT_CALLABLE_DECLARE(int inet6_recvmsg(struct socket *, struct msghdr *, | |
939 | size_t, int)); | |
1b784140 | 940 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 941 | int flags) |
1da177e4 | 942 | { |
a648a592 PA |
943 | return INDIRECT_CALL_INET(sock->ops->recvmsg, inet6_recvmsg, |
944 | inet_recvmsg, sock, msg, msg_data_left(msg), | |
945 | flags); | |
1da177e4 LT |
946 | } |
947 | ||
85806af0 RD |
948 | /** |
949 | * sock_recvmsg - receive a message from @sock | |
950 | * @sock: socket | |
951 | * @msg: message to receive | |
952 | * @flags: message flags | |
953 | * | |
954 | * Receives @msg from @sock, passing through LSM. Returns the total number | |
955 | * of bytes received, or an error. | |
956 | */ | |
2da62906 | 957 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 958 | { |
2da62906 | 959 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 960 | |
2da62906 | 961 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 962 | } |
c6d409cf | 963 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 964 | |
c1249c0a | 965 | /** |
8a3c245c PT |
966 | * kernel_recvmsg - Receive a message from a socket (kernel space) |
967 | * @sock: The socket to receive the message from | |
968 | * @msg: Received message | |
969 | * @vec: Input s/g array for message data | |
970 | * @num: Size of input s/g array | |
971 | * @size: Number of bytes to read | |
972 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
c1249c0a | 973 | * |
8a3c245c PT |
974 | * On return the msg structure contains the scatter/gather array passed in the |
975 | * vec argument. The array is modified so that it consists of the unfilled | |
976 | * portion of the original array. | |
c1249c0a | 977 | * |
8a3c245c | 978 | * The returned value is the total number of bytes received, or an error. |
c1249c0a | 979 | */ |
8a3c245c | 980 | |
89bddce5 SH |
981 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
982 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 | 983 | { |
1f466e1f | 984 | msg->msg_control_is_user = false; |
aa563d7b | 985 | iov_iter_kvec(&msg->msg_iter, READ, vec, num, size); |
1f466e1f | 986 | return sock_recvmsg(sock, msg, flags); |
1da177e4 | 987 | } |
c6d409cf | 988 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 989 | |
ce1d4d3e CH |
990 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
991 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 992 | { |
1da177e4 LT |
993 | struct socket *sock; |
994 | int flags; | |
995 | ||
ce1d4d3e CH |
996 | sock = file->private_data; |
997 | ||
35f9c09f ED |
998 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
999 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
1000 | flags |= more; | |
ce1d4d3e | 1001 | |
e6949583 | 1002 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 1003 | } |
1da177e4 | 1004 | |
9c55e01c | 1005 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 1006 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
1007 | unsigned int flags) |
1008 | { | |
1009 | struct socket *sock = file->private_data; | |
1010 | ||
997b37da | 1011 | if (unlikely(!sock->ops->splice_read)) |
95506588 | 1012 | return generic_file_splice_read(file, ppos, pipe, len, flags); |
997b37da | 1013 | |
9c55e01c JA |
1014 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
1015 | } | |
1016 | ||
8ae5e030 | 1017 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 1018 | { |
6d652330 AV |
1019 | struct file *file = iocb->ki_filp; |
1020 | struct socket *sock = file->private_data; | |
0345f931 | 1021 | struct msghdr msg = {.msg_iter = *to, |
1022 | .msg_iocb = iocb}; | |
8ae5e030 | 1023 | ssize_t res; |
ce1d4d3e | 1024 | |
ebfcd895 | 1025 | if (file->f_flags & O_NONBLOCK || (iocb->ki_flags & IOCB_NOWAIT)) |
8ae5e030 AV |
1026 | msg.msg_flags = MSG_DONTWAIT; |
1027 | ||
1028 | if (iocb->ki_pos != 0) | |
1da177e4 | 1029 | return -ESPIPE; |
027445c3 | 1030 | |
66ee59af | 1031 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
1032 | return 0; |
1033 | ||
2da62906 | 1034 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
1035 | *to = msg.msg_iter; |
1036 | return res; | |
1da177e4 LT |
1037 | } |
1038 | ||
8ae5e030 | 1039 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 1040 | { |
6d652330 AV |
1041 | struct file *file = iocb->ki_filp; |
1042 | struct socket *sock = file->private_data; | |
0345f931 | 1043 | struct msghdr msg = {.msg_iter = *from, |
1044 | .msg_iocb = iocb}; | |
8ae5e030 | 1045 | ssize_t res; |
1da177e4 | 1046 | |
8ae5e030 | 1047 | if (iocb->ki_pos != 0) |
ce1d4d3e | 1048 | return -ESPIPE; |
027445c3 | 1049 | |
ebfcd895 | 1050 | if (file->f_flags & O_NONBLOCK || (iocb->ki_flags & IOCB_NOWAIT)) |
8ae5e030 AV |
1051 | msg.msg_flags = MSG_DONTWAIT; |
1052 | ||
6d652330 AV |
1053 | if (sock->type == SOCK_SEQPACKET) |
1054 | msg.msg_flags |= MSG_EOR; | |
1055 | ||
d8725c86 | 1056 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
1057 | *from = msg.msg_iter; |
1058 | return res; | |
1da177e4 LT |
1059 | } |
1060 | ||
1da177e4 LT |
1061 | /* |
1062 | * Atomic setting of ioctl hooks to avoid race | |
1063 | * with module unload. | |
1064 | */ | |
1065 | ||
4a3e2f71 | 1066 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 1067 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 1068 | |
881d966b | 1069 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 1070 | { |
4a3e2f71 | 1071 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 1072 | br_ioctl_hook = hook; |
4a3e2f71 | 1073 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1074 | } |
1075 | EXPORT_SYMBOL(brioctl_set); | |
1076 | ||
4a3e2f71 | 1077 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 1078 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 1079 | |
881d966b | 1080 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 1081 | { |
4a3e2f71 | 1082 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1083 | vlan_ioctl_hook = hook; |
4a3e2f71 | 1084 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
1085 | } |
1086 | EXPORT_SYMBOL(vlan_ioctl_set); | |
1087 | ||
6b96018b | 1088 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
63ff03ab | 1089 | unsigned int cmd, unsigned long arg) |
6b96018b | 1090 | { |
876f0bf9 AB |
1091 | struct ifreq ifr; |
1092 | bool need_copyout; | |
6b96018b AB |
1093 | int err; |
1094 | void __user *argp = (void __user *)arg; | |
1095 | ||
1096 | err = sock->ops->ioctl(sock, cmd, arg); | |
1097 | ||
1098 | /* | |
1099 | * If this ioctl is unknown try to hand it down | |
1100 | * to the NIC driver. | |
1101 | */ | |
36fd633e AV |
1102 | if (err != -ENOIOCTLCMD) |
1103 | return err; | |
6b96018b | 1104 | |
876f0bf9 AB |
1105 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) |
1106 | return -EFAULT; | |
1107 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1108 | if (!err && need_copyout) | |
1109 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
44c02a2c | 1110 | return -EFAULT; |
876f0bf9 | 1111 | |
6b96018b AB |
1112 | return err; |
1113 | } | |
1114 | ||
1da177e4 LT |
1115 | /* |
1116 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1117 | * what to do with it - that's up to the protocol still. | |
1118 | */ | |
1119 | ||
1120 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) | |
1121 | { | |
1122 | struct socket *sock; | |
881d966b | 1123 | struct sock *sk; |
1da177e4 LT |
1124 | void __user *argp = (void __user *)arg; |
1125 | int pid, err; | |
881d966b | 1126 | struct net *net; |
1da177e4 | 1127 | |
b69aee04 | 1128 | sock = file->private_data; |
881d966b | 1129 | sk = sock->sk; |
3b1e0a65 | 1130 | net = sock_net(sk); |
44c02a2c AV |
1131 | if (unlikely(cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))) { |
1132 | struct ifreq ifr; | |
1133 | bool need_copyout; | |
1134 | if (copy_from_user(&ifr, argp, sizeof(struct ifreq))) | |
1135 | return -EFAULT; | |
1136 | err = dev_ioctl(net, cmd, &ifr, &need_copyout); | |
1137 | if (!err && need_copyout) | |
1138 | if (copy_to_user(argp, &ifr, sizeof(struct ifreq))) | |
1139 | return -EFAULT; | |
1da177e4 | 1140 | } else |
3d23e349 | 1141 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1142 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
b1b0c245 | 1143 | err = wext_handle_ioctl(net, cmd, argp); |
1da177e4 | 1144 | } else |
3d23e349 | 1145 | #endif |
89bddce5 | 1146 | switch (cmd) { |
1da177e4 LT |
1147 | case FIOSETOWN: |
1148 | case SIOCSPGRP: | |
1149 | err = -EFAULT; | |
1150 | if (get_user(pid, (int __user *)argp)) | |
1151 | break; | |
393cc3f5 | 1152 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1153 | break; |
1154 | case FIOGETOWN: | |
1155 | case SIOCGPGRP: | |
609d7fa9 | 1156 | err = put_user(f_getown(sock->file), |
89bddce5 | 1157 | (int __user *)argp); |
1da177e4 LT |
1158 | break; |
1159 | case SIOCGIFBR: | |
1160 | case SIOCSIFBR: | |
1161 | case SIOCBRADDBR: | |
1162 | case SIOCBRDELBR: | |
1163 | err = -ENOPKG; | |
1164 | if (!br_ioctl_hook) | |
1165 | request_module("bridge"); | |
1166 | ||
4a3e2f71 | 1167 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1168 | if (br_ioctl_hook) |
881d966b | 1169 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1170 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1171 | break; |
1172 | case SIOCGIFVLAN: | |
1173 | case SIOCSIFVLAN: | |
1174 | err = -ENOPKG; | |
1175 | if (!vlan_ioctl_hook) | |
1176 | request_module("8021q"); | |
1177 | ||
4a3e2f71 | 1178 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1179 | if (vlan_ioctl_hook) |
881d966b | 1180 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1181 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1182 | break; |
c62cce2c AV |
1183 | case SIOCGSKNS: |
1184 | err = -EPERM; | |
1185 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1186 | break; | |
1187 | ||
1188 | err = open_related_ns(&net->ns, get_net_ns); | |
1189 | break; | |
0768e170 AB |
1190 | case SIOCGSTAMP_OLD: |
1191 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
1192 | if (!sock->ops->gettstamp) { |
1193 | err = -ENOIOCTLCMD; | |
1194 | break; | |
1195 | } | |
1196 | err = sock->ops->gettstamp(sock, argp, | |
0768e170 AB |
1197 | cmd == SIOCGSTAMP_OLD, |
1198 | !IS_ENABLED(CONFIG_64BIT)); | |
60747828 | 1199 | break; |
0768e170 AB |
1200 | case SIOCGSTAMP_NEW: |
1201 | case SIOCGSTAMPNS_NEW: | |
1202 | if (!sock->ops->gettstamp) { | |
1203 | err = -ENOIOCTLCMD; | |
1204 | break; | |
1205 | } | |
1206 | err = sock->ops->gettstamp(sock, argp, | |
1207 | cmd == SIOCGSTAMP_NEW, | |
1208 | false); | |
c7cbdbf2 | 1209 | break; |
876f0bf9 AB |
1210 | |
1211 | case SIOCGIFCONF: | |
1212 | err = dev_ifconf(net, argp); | |
1213 | break; | |
1214 | ||
1da177e4 | 1215 | default: |
63ff03ab | 1216 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1217 | break; |
89bddce5 | 1218 | } |
1da177e4 LT |
1219 | return err; |
1220 | } | |
1221 | ||
8a3c245c PT |
1222 | /** |
1223 | * sock_create_lite - creates a socket | |
1224 | * @family: protocol family (AF_INET, ...) | |
1225 | * @type: communication type (SOCK_STREAM, ...) | |
1226 | * @protocol: protocol (0, ...) | |
1227 | * @res: new socket | |
1228 | * | |
1229 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1230 | * The new socket initialization is not complete, see kernel_accept(). | |
1231 | * Returns 0 or an error. On failure @res is set to %NULL. | |
1232 | * This function internally uses GFP_KERNEL. | |
1233 | */ | |
1234 | ||
1da177e4 LT |
1235 | int sock_create_lite(int family, int type, int protocol, struct socket **res) |
1236 | { | |
1237 | int err; | |
1238 | struct socket *sock = NULL; | |
89bddce5 | 1239 | |
1da177e4 LT |
1240 | err = security_socket_create(family, type, protocol, 1); |
1241 | if (err) | |
1242 | goto out; | |
1243 | ||
1244 | sock = sock_alloc(); | |
1245 | if (!sock) { | |
1246 | err = -ENOMEM; | |
1247 | goto out; | |
1248 | } | |
1249 | ||
1da177e4 | 1250 | sock->type = type; |
7420ed23 VY |
1251 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1252 | if (err) | |
1253 | goto out_release; | |
1254 | ||
1da177e4 LT |
1255 | out: |
1256 | *res = sock; | |
1257 | return err; | |
7420ed23 VY |
1258 | out_release: |
1259 | sock_release(sock); | |
1260 | sock = NULL; | |
1261 | goto out; | |
1da177e4 | 1262 | } |
c6d409cf | 1263 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1264 | |
1265 | /* No kernel lock held - perfect */ | |
ade994f4 | 1266 | static __poll_t sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1267 | { |
3cafb376 | 1268 | struct socket *sock = file->private_data; |
a331de3b | 1269 | __poll_t events = poll_requested_events(wait), flag = 0; |
2d48d67f | 1270 | |
e88958e6 CH |
1271 | if (!sock->ops->poll) |
1272 | return 0; | |
f641f13b | 1273 | |
a331de3b CH |
1274 | if (sk_can_busy_loop(sock->sk)) { |
1275 | /* poll once if requested by the syscall */ | |
1276 | if (events & POLL_BUSY_LOOP) | |
1277 | sk_busy_loop(sock->sk, 1); | |
1278 | ||
1279 | /* if this socket can poll_ll, tell the system call */ | |
1280 | flag = POLL_BUSY_LOOP; | |
1281 | } | |
1282 | ||
1283 | return sock->ops->poll(file, sock, wait) | flag; | |
1da177e4 LT |
1284 | } |
1285 | ||
89bddce5 | 1286 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1287 | { |
b69aee04 | 1288 | struct socket *sock = file->private_data; |
1da177e4 LT |
1289 | |
1290 | return sock->ops->mmap(file, sock, vma); | |
1291 | } | |
1292 | ||
20380731 | 1293 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1294 | { |
6d8c50dc | 1295 | __sock_release(SOCKET_I(inode), inode); |
1da177e4 LT |
1296 | return 0; |
1297 | } | |
1298 | ||
1299 | /* | |
1300 | * Update the socket async list | |
1301 | * | |
1302 | * Fasync_list locking strategy. | |
1303 | * | |
1304 | * 1. fasync_list is modified only under process context socket lock | |
1305 | * i.e. under semaphore. | |
1306 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1307 | * or under socket lock |
1da177e4 LT |
1308 | */ |
1309 | ||
1310 | static int sock_fasync(int fd, struct file *filp, int on) | |
1311 | { | |
989a2979 ED |
1312 | struct socket *sock = filp->private_data; |
1313 | struct sock *sk = sock->sk; | |
333f7909 | 1314 | struct socket_wq *wq = &sock->wq; |
1da177e4 | 1315 | |
989a2979 | 1316 | if (sk == NULL) |
1da177e4 | 1317 | return -EINVAL; |
1da177e4 LT |
1318 | |
1319 | lock_sock(sk); | |
eaefd110 | 1320 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1321 | |
eaefd110 | 1322 | if (!wq->fasync_list) |
989a2979 ED |
1323 | sock_reset_flag(sk, SOCK_FASYNC); |
1324 | else | |
bcdce719 | 1325 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1326 | |
989a2979 | 1327 | release_sock(sk); |
1da177e4 LT |
1328 | return 0; |
1329 | } | |
1330 | ||
ceb5d58b | 1331 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1332 | |
ceb5d58b | 1333 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1334 | { |
ceb5d58b | 1335 | if (!wq || !wq->fasync_list) |
1da177e4 | 1336 | return -1; |
ceb5d58b | 1337 | |
89bddce5 | 1338 | switch (how) { |
8d8ad9d7 | 1339 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1340 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1341 | break; |
1342 | goto call_kill; | |
8d8ad9d7 | 1343 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1344 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 | 1345 | break; |
7c7ab580 | 1346 | fallthrough; |
8d8ad9d7 | 1347 | case SOCK_WAKE_IO: |
89bddce5 | 1348 | call_kill: |
43815482 | 1349 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1350 | break; |
8d8ad9d7 | 1351 | case SOCK_WAKE_URG: |
43815482 | 1352 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1353 | } |
ceb5d58b | 1354 | |
1da177e4 LT |
1355 | return 0; |
1356 | } | |
c6d409cf | 1357 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1358 | |
8a3c245c PT |
1359 | /** |
1360 | * __sock_create - creates a socket | |
1361 | * @net: net namespace | |
1362 | * @family: protocol family (AF_INET, ...) | |
1363 | * @type: communication type (SOCK_STREAM, ...) | |
1364 | * @protocol: protocol (0, ...) | |
1365 | * @res: new socket | |
1366 | * @kern: boolean for kernel space sockets | |
1367 | * | |
1368 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1369 | * Returns 0 or an error. On failure @res is set to %NULL. @kern must | |
1370 | * be set to true if the socket resides in kernel space. | |
1371 | * This function internally uses GFP_KERNEL. | |
1372 | */ | |
1373 | ||
721db93a | 1374 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1375 | struct socket **res, int kern) |
1da177e4 LT |
1376 | { |
1377 | int err; | |
1378 | struct socket *sock; | |
55737fda | 1379 | const struct net_proto_family *pf; |
1da177e4 LT |
1380 | |
1381 | /* | |
89bddce5 | 1382 | * Check protocol is in range |
1da177e4 LT |
1383 | */ |
1384 | if (family < 0 || family >= NPROTO) | |
1385 | return -EAFNOSUPPORT; | |
1386 | if (type < 0 || type >= SOCK_MAX) | |
1387 | return -EINVAL; | |
1388 | ||
1389 | /* Compatibility. | |
1390 | ||
1391 | This uglymoron is moved from INET layer to here to avoid | |
1392 | deadlock in module load. | |
1393 | */ | |
1394 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1395 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1396 | current->comm); | |
1da177e4 LT |
1397 | family = PF_PACKET; |
1398 | } | |
1399 | ||
1400 | err = security_socket_create(family, type, protocol, kern); | |
1401 | if (err) | |
1402 | return err; | |
89bddce5 | 1403 | |
55737fda SH |
1404 | /* |
1405 | * Allocate the socket and allow the family to set things up. if | |
1406 | * the protocol is 0, the family is instructed to select an appropriate | |
1407 | * default. | |
1408 | */ | |
1409 | sock = sock_alloc(); | |
1410 | if (!sock) { | |
e87cc472 | 1411 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1412 | return -ENFILE; /* Not exactly a match, but its the |
1413 | closest posix thing */ | |
1414 | } | |
1415 | ||
1416 | sock->type = type; | |
1417 | ||
95a5afca | 1418 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1419 | /* Attempt to load a protocol module if the find failed. |
1420 | * | |
1421 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1422 | * requested real, full-featured networking support upon configuration. |
1423 | * Otherwise module support will break! | |
1424 | */ | |
190683a9 | 1425 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1426 | request_module("net-pf-%d", family); |
1da177e4 LT |
1427 | #endif |
1428 | ||
55737fda SH |
1429 | rcu_read_lock(); |
1430 | pf = rcu_dereference(net_families[family]); | |
1431 | err = -EAFNOSUPPORT; | |
1432 | if (!pf) | |
1433 | goto out_release; | |
1da177e4 LT |
1434 | |
1435 | /* | |
1436 | * We will call the ->create function, that possibly is in a loadable | |
1437 | * module, so we have to bump that loadable module refcnt first. | |
1438 | */ | |
55737fda | 1439 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1440 | goto out_release; |
1441 | ||
55737fda SH |
1442 | /* Now protected by module ref count */ |
1443 | rcu_read_unlock(); | |
1444 | ||
3f378b68 | 1445 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1446 | if (err < 0) |
1da177e4 | 1447 | goto out_module_put; |
a79af59e | 1448 | |
1da177e4 LT |
1449 | /* |
1450 | * Now to bump the refcnt of the [loadable] module that owns this | |
1451 | * socket at sock_release time we decrement its refcnt. | |
1452 | */ | |
55737fda SH |
1453 | if (!try_module_get(sock->ops->owner)) |
1454 | goto out_module_busy; | |
1455 | ||
1da177e4 LT |
1456 | /* |
1457 | * Now that we're done with the ->create function, the [loadable] | |
1458 | * module can have its refcnt decremented | |
1459 | */ | |
55737fda | 1460 | module_put(pf->owner); |
7420ed23 VY |
1461 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1462 | if (err) | |
3b185525 | 1463 | goto out_sock_release; |
55737fda | 1464 | *res = sock; |
1da177e4 | 1465 | |
55737fda SH |
1466 | return 0; |
1467 | ||
1468 | out_module_busy: | |
1469 | err = -EAFNOSUPPORT; | |
1da177e4 | 1470 | out_module_put: |
55737fda SH |
1471 | sock->ops = NULL; |
1472 | module_put(pf->owner); | |
1473 | out_sock_release: | |
1da177e4 | 1474 | sock_release(sock); |
55737fda SH |
1475 | return err; |
1476 | ||
1477 | out_release: | |
1478 | rcu_read_unlock(); | |
1479 | goto out_sock_release; | |
1da177e4 | 1480 | } |
721db93a | 1481 | EXPORT_SYMBOL(__sock_create); |
1da177e4 | 1482 | |
8a3c245c PT |
1483 | /** |
1484 | * sock_create - creates a socket | |
1485 | * @family: protocol family (AF_INET, ...) | |
1486 | * @type: communication type (SOCK_STREAM, ...) | |
1487 | * @protocol: protocol (0, ...) | |
1488 | * @res: new socket | |
1489 | * | |
1490 | * A wrapper around __sock_create(). | |
1491 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1492 | */ | |
1493 | ||
1da177e4 LT |
1494 | int sock_create(int family, int type, int protocol, struct socket **res) |
1495 | { | |
1b8d7ae4 | 1496 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1497 | } |
c6d409cf | 1498 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1499 | |
8a3c245c PT |
1500 | /** |
1501 | * sock_create_kern - creates a socket (kernel space) | |
1502 | * @net: net namespace | |
1503 | * @family: protocol family (AF_INET, ...) | |
1504 | * @type: communication type (SOCK_STREAM, ...) | |
1505 | * @protocol: protocol (0, ...) | |
1506 | * @res: new socket | |
1507 | * | |
1508 | * A wrapper around __sock_create(). | |
1509 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1510 | */ | |
1511 | ||
eeb1bd5c | 1512 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1513 | { |
eeb1bd5c | 1514 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1515 | } |
c6d409cf | 1516 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1517 | |
9d6a15c3 | 1518 | int __sys_socket(int family, int type, int protocol) |
1da177e4 LT |
1519 | { |
1520 | int retval; | |
1521 | struct socket *sock; | |
a677a039 UD |
1522 | int flags; |
1523 | ||
e38b36f3 UD |
1524 | /* Check the SOCK_* constants for consistency. */ |
1525 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1526 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1527 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1528 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1529 | ||
a677a039 | 1530 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1531 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1532 | return -EINVAL; |
1533 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1534 | |
aaca0bdc UD |
1535 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1536 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1537 | ||
1da177e4 LT |
1538 | retval = sock_create(family, type, protocol, &sock); |
1539 | if (retval < 0) | |
8e1611e2 | 1540 | return retval; |
1da177e4 | 1541 | |
8e1611e2 | 1542 | return sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1543 | } |
1544 | ||
9d6a15c3 DB |
1545 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1546 | { | |
1547 | return __sys_socket(family, type, protocol); | |
1548 | } | |
1549 | ||
1da177e4 LT |
1550 | /* |
1551 | * Create a pair of connected sockets. | |
1552 | */ | |
1553 | ||
6debc8d8 | 1554 | int __sys_socketpair(int family, int type, int protocol, int __user *usockvec) |
1da177e4 LT |
1555 | { |
1556 | struct socket *sock1, *sock2; | |
1557 | int fd1, fd2, err; | |
db349509 | 1558 | struct file *newfile1, *newfile2; |
a677a039 UD |
1559 | int flags; |
1560 | ||
1561 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1562 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1563 | return -EINVAL; |
1564 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1565 | |
aaca0bdc UD |
1566 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1567 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1568 | ||
016a266b AV |
1569 | /* |
1570 | * reserve descriptors and make sure we won't fail | |
1571 | * to return them to userland. | |
1572 | */ | |
1573 | fd1 = get_unused_fd_flags(flags); | |
1574 | if (unlikely(fd1 < 0)) | |
1575 | return fd1; | |
1576 | ||
1577 | fd2 = get_unused_fd_flags(flags); | |
1578 | if (unlikely(fd2 < 0)) { | |
1579 | put_unused_fd(fd1); | |
1580 | return fd2; | |
1581 | } | |
1582 | ||
1583 | err = put_user(fd1, &usockvec[0]); | |
1584 | if (err) | |
1585 | goto out; | |
1586 | ||
1587 | err = put_user(fd2, &usockvec[1]); | |
1588 | if (err) | |
1589 | goto out; | |
1590 | ||
1da177e4 LT |
1591 | /* |
1592 | * Obtain the first socket and check if the underlying protocol | |
1593 | * supports the socketpair call. | |
1594 | */ | |
1595 | ||
1596 | err = sock_create(family, type, protocol, &sock1); | |
016a266b | 1597 | if (unlikely(err < 0)) |
1da177e4 LT |
1598 | goto out; |
1599 | ||
1600 | err = sock_create(family, type, protocol, &sock2); | |
016a266b AV |
1601 | if (unlikely(err < 0)) { |
1602 | sock_release(sock1); | |
1603 | goto out; | |
bf3c23d1 | 1604 | } |
d73aa286 | 1605 | |
d47cd945 DH |
1606 | err = security_socket_socketpair(sock1, sock2); |
1607 | if (unlikely(err)) { | |
1608 | sock_release(sock2); | |
1609 | sock_release(sock1); | |
1610 | goto out; | |
1611 | } | |
1612 | ||
016a266b AV |
1613 | err = sock1->ops->socketpair(sock1, sock2); |
1614 | if (unlikely(err < 0)) { | |
1615 | sock_release(sock2); | |
1616 | sock_release(sock1); | |
1617 | goto out; | |
28407630 AV |
1618 | } |
1619 | ||
aab174f0 | 1620 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1621 | if (IS_ERR(newfile1)) { |
28407630 | 1622 | err = PTR_ERR(newfile1); |
016a266b AV |
1623 | sock_release(sock2); |
1624 | goto out; | |
28407630 AV |
1625 | } |
1626 | ||
aab174f0 | 1627 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1628 | if (IS_ERR(newfile2)) { |
1629 | err = PTR_ERR(newfile2); | |
016a266b AV |
1630 | fput(newfile1); |
1631 | goto out; | |
db349509 AV |
1632 | } |
1633 | ||
157cf649 | 1634 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1635 | |
db349509 AV |
1636 | fd_install(fd1, newfile1); |
1637 | fd_install(fd2, newfile2); | |
d73aa286 | 1638 | return 0; |
1da177e4 | 1639 | |
016a266b | 1640 | out: |
d73aa286 | 1641 | put_unused_fd(fd2); |
d73aa286 | 1642 | put_unused_fd(fd1); |
1da177e4 LT |
1643 | return err; |
1644 | } | |
1645 | ||
6debc8d8 DB |
1646 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1647 | int __user *, usockvec) | |
1648 | { | |
1649 | return __sys_socketpair(family, type, protocol, usockvec); | |
1650 | } | |
1651 | ||
1da177e4 LT |
1652 | /* |
1653 | * Bind a name to a socket. Nothing much to do here since it's | |
1654 | * the protocol's responsibility to handle the local address. | |
1655 | * | |
1656 | * We move the socket address to kernel space before we call | |
1657 | * the protocol layer (having also checked the address is ok). | |
1658 | */ | |
1659 | ||
a87d35d8 | 1660 | int __sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
1da177e4 LT |
1661 | { |
1662 | struct socket *sock; | |
230b1839 | 1663 | struct sockaddr_storage address; |
6cb153ca | 1664 | int err, fput_needed; |
1da177e4 | 1665 | |
89bddce5 | 1666 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1667 | if (sock) { |
43db362d | 1668 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
068b88cc | 1669 | if (!err) { |
89bddce5 | 1670 | err = security_socket_bind(sock, |
230b1839 | 1671 | (struct sockaddr *)&address, |
89bddce5 | 1672 | addrlen); |
6cb153ca BL |
1673 | if (!err) |
1674 | err = sock->ops->bind(sock, | |
89bddce5 | 1675 | (struct sockaddr *) |
230b1839 | 1676 | &address, addrlen); |
1da177e4 | 1677 | } |
6cb153ca | 1678 | fput_light(sock->file, fput_needed); |
89bddce5 | 1679 | } |
1da177e4 LT |
1680 | return err; |
1681 | } | |
1682 | ||
a87d35d8 DB |
1683 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1684 | { | |
1685 | return __sys_bind(fd, umyaddr, addrlen); | |
1686 | } | |
1687 | ||
1da177e4 LT |
1688 | /* |
1689 | * Perform a listen. Basically, we allow the protocol to do anything | |
1690 | * necessary for a listen, and if that works, we mark the socket as | |
1691 | * ready for listening. | |
1692 | */ | |
1693 | ||
25e290ee | 1694 | int __sys_listen(int fd, int backlog) |
1da177e4 LT |
1695 | { |
1696 | struct socket *sock; | |
6cb153ca | 1697 | int err, fput_needed; |
b8e1f9b5 | 1698 | int somaxconn; |
89bddce5 SH |
1699 | |
1700 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1701 | if (sock) { | |
8efa6e93 | 1702 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1703 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1704 | backlog = somaxconn; |
1da177e4 LT |
1705 | |
1706 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1707 | if (!err) |
1708 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1709 | |
6cb153ca | 1710 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1711 | } |
1712 | return err; | |
1713 | } | |
1714 | ||
25e290ee DB |
1715 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1716 | { | |
1717 | return __sys_listen(fd, backlog); | |
1718 | } | |
1719 | ||
de2ea4b6 JA |
1720 | int __sys_accept4_file(struct file *file, unsigned file_flags, |
1721 | struct sockaddr __user *upeer_sockaddr, | |
09952e3e JA |
1722 | int __user *upeer_addrlen, int flags, |
1723 | unsigned long nofile) | |
1da177e4 LT |
1724 | { |
1725 | struct socket *sock, *newsock; | |
39d8c1b6 | 1726 | struct file *newfile; |
de2ea4b6 | 1727 | int err, len, newfd; |
230b1839 | 1728 | struct sockaddr_storage address; |
1da177e4 | 1729 | |
77d27200 | 1730 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1731 | return -EINVAL; |
1732 | ||
1733 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1734 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1735 | ||
dba4a925 FR |
1736 | sock = sock_from_file(file); |
1737 | if (!sock) { | |
1738 | err = -ENOTSOCK; | |
1da177e4 | 1739 | goto out; |
dba4a925 | 1740 | } |
1da177e4 LT |
1741 | |
1742 | err = -ENFILE; | |
c6d409cf ED |
1743 | newsock = sock_alloc(); |
1744 | if (!newsock) | |
de2ea4b6 | 1745 | goto out; |
1da177e4 LT |
1746 | |
1747 | newsock->type = sock->type; | |
1748 | newsock->ops = sock->ops; | |
1749 | ||
1da177e4 LT |
1750 | /* |
1751 | * We don't need try_module_get here, as the listening socket (sock) | |
1752 | * has the protocol module (sock->ops->owner) held. | |
1753 | */ | |
1754 | __module_get(newsock->ops->owner); | |
1755 | ||
09952e3e | 1756 | newfd = __get_unused_fd_flags(flags, nofile); |
39d8c1b6 DM |
1757 | if (unlikely(newfd < 0)) { |
1758 | err = newfd; | |
9a1875e6 | 1759 | sock_release(newsock); |
de2ea4b6 | 1760 | goto out; |
39d8c1b6 | 1761 | } |
aab174f0 | 1762 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1763 | if (IS_ERR(newfile)) { |
28407630 AV |
1764 | err = PTR_ERR(newfile); |
1765 | put_unused_fd(newfd); | |
de2ea4b6 | 1766 | goto out; |
28407630 | 1767 | } |
39d8c1b6 | 1768 | |
a79af59e FF |
1769 | err = security_socket_accept(sock, newsock); |
1770 | if (err) | |
39d8c1b6 | 1771 | goto out_fd; |
a79af59e | 1772 | |
de2ea4b6 JA |
1773 | err = sock->ops->accept(sock, newsock, sock->file->f_flags | file_flags, |
1774 | false); | |
1da177e4 | 1775 | if (err < 0) |
39d8c1b6 | 1776 | goto out_fd; |
1da177e4 LT |
1777 | |
1778 | if (upeer_sockaddr) { | |
9b2c45d4 DV |
1779 | len = newsock->ops->getname(newsock, |
1780 | (struct sockaddr *)&address, 2); | |
1781 | if (len < 0) { | |
1da177e4 | 1782 | err = -ECONNABORTED; |
39d8c1b6 | 1783 | goto out_fd; |
1da177e4 | 1784 | } |
43db362d | 1785 | err = move_addr_to_user(&address, |
230b1839 | 1786 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1787 | if (err < 0) |
39d8c1b6 | 1788 | goto out_fd; |
1da177e4 LT |
1789 | } |
1790 | ||
1791 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1792 | ||
39d8c1b6 DM |
1793 | fd_install(newfd, newfile); |
1794 | err = newfd; | |
1da177e4 LT |
1795 | out: |
1796 | return err; | |
39d8c1b6 | 1797 | out_fd: |
9606a216 | 1798 | fput(newfile); |
39d8c1b6 | 1799 | put_unused_fd(newfd); |
de2ea4b6 JA |
1800 | goto out; |
1801 | ||
1802 | } | |
1803 | ||
1804 | /* | |
1805 | * For accept, we attempt to create a new socket, set up the link | |
1806 | * with the client, wake up the client, then return the new | |
1807 | * connected fd. We collect the address of the connector in kernel | |
1808 | * space and move it to user at the very end. This is unclean because | |
1809 | * we open the socket then return an error. | |
1810 | * | |
1811 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1812 | * status to recvmsg. We need to add that support in a way thats | |
1813 | * clean when we restructure accept also. | |
1814 | */ | |
1815 | ||
1816 | int __sys_accept4(int fd, struct sockaddr __user *upeer_sockaddr, | |
1817 | int __user *upeer_addrlen, int flags) | |
1818 | { | |
1819 | int ret = -EBADF; | |
1820 | struct fd f; | |
1821 | ||
1822 | f = fdget(fd); | |
1823 | if (f.file) { | |
1824 | ret = __sys_accept4_file(f.file, 0, upeer_sockaddr, | |
09952e3e JA |
1825 | upeer_addrlen, flags, |
1826 | rlimit(RLIMIT_NOFILE)); | |
6b07edeb | 1827 | fdput(f); |
de2ea4b6 JA |
1828 | } |
1829 | ||
1830 | return ret; | |
1da177e4 LT |
1831 | } |
1832 | ||
4541e805 DB |
1833 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1834 | int __user *, upeer_addrlen, int, flags) | |
1835 | { | |
1836 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, flags); | |
1837 | } | |
1838 | ||
20f37034 HC |
1839 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1840 | int __user *, upeer_addrlen) | |
aaca0bdc | 1841 | { |
4541e805 | 1842 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1843 | } |
1844 | ||
1da177e4 LT |
1845 | /* |
1846 | * Attempt to connect to a socket with the server address. The address | |
1847 | * is in user space so we verify it is OK and move it to kernel space. | |
1848 | * | |
1849 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1850 | * break bindings | |
1851 | * | |
1852 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1853 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1854 | * include the -EINPROGRESS status for such sockets. | |
1855 | */ | |
1856 | ||
f499a021 | 1857 | int __sys_connect_file(struct file *file, struct sockaddr_storage *address, |
bd3ded31 | 1858 | int addrlen, int file_flags) |
1da177e4 LT |
1859 | { |
1860 | struct socket *sock; | |
bd3ded31 | 1861 | int err; |
1da177e4 | 1862 | |
dba4a925 FR |
1863 | sock = sock_from_file(file); |
1864 | if (!sock) { | |
1865 | err = -ENOTSOCK; | |
1da177e4 | 1866 | goto out; |
dba4a925 | 1867 | } |
1da177e4 | 1868 | |
89bddce5 | 1869 | err = |
f499a021 | 1870 | security_socket_connect(sock, (struct sockaddr *)address, addrlen); |
1da177e4 | 1871 | if (err) |
bd3ded31 | 1872 | goto out; |
1da177e4 | 1873 | |
f499a021 | 1874 | err = sock->ops->connect(sock, (struct sockaddr *)address, addrlen, |
bd3ded31 | 1875 | sock->file->f_flags | file_flags); |
1da177e4 LT |
1876 | out: |
1877 | return err; | |
1878 | } | |
1879 | ||
bd3ded31 JA |
1880 | int __sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
1881 | { | |
1882 | int ret = -EBADF; | |
1883 | struct fd f; | |
1884 | ||
1885 | f = fdget(fd); | |
1886 | if (f.file) { | |
f499a021 JA |
1887 | struct sockaddr_storage address; |
1888 | ||
1889 | ret = move_addr_to_kernel(uservaddr, addrlen, &address); | |
1890 | if (!ret) | |
1891 | ret = __sys_connect_file(f.file, &address, addrlen, 0); | |
6b07edeb | 1892 | fdput(f); |
bd3ded31 JA |
1893 | } |
1894 | ||
1895 | return ret; | |
1896 | } | |
1897 | ||
1387c2c2 DB |
1898 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1899 | int, addrlen) | |
1900 | { | |
1901 | return __sys_connect(fd, uservaddr, addrlen); | |
1902 | } | |
1903 | ||
1da177e4 LT |
1904 | /* |
1905 | * Get the local address ('name') of a socket object. Move the obtained | |
1906 | * name to user space. | |
1907 | */ | |
1908 | ||
8882a107 DB |
1909 | int __sys_getsockname(int fd, struct sockaddr __user *usockaddr, |
1910 | int __user *usockaddr_len) | |
1da177e4 LT |
1911 | { |
1912 | struct socket *sock; | |
230b1839 | 1913 | struct sockaddr_storage address; |
9b2c45d4 | 1914 | int err, fput_needed; |
89bddce5 | 1915 | |
6cb153ca | 1916 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1917 | if (!sock) |
1918 | goto out; | |
1919 | ||
1920 | err = security_socket_getsockname(sock); | |
1921 | if (err) | |
1922 | goto out_put; | |
1923 | ||
9b2c45d4 DV |
1924 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 0); |
1925 | if (err < 0) | |
1da177e4 | 1926 | goto out_put; |
9b2c45d4 DV |
1927 | /* "err" is actually length in this case */ |
1928 | err = move_addr_to_user(&address, err, usockaddr, usockaddr_len); | |
1da177e4 LT |
1929 | |
1930 | out_put: | |
6cb153ca | 1931 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1932 | out: |
1933 | return err; | |
1934 | } | |
1935 | ||
8882a107 DB |
1936 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1937 | int __user *, usockaddr_len) | |
1938 | { | |
1939 | return __sys_getsockname(fd, usockaddr, usockaddr_len); | |
1940 | } | |
1941 | ||
1da177e4 LT |
1942 | /* |
1943 | * Get the remote address ('name') of a socket object. Move the obtained | |
1944 | * name to user space. | |
1945 | */ | |
1946 | ||
b21c8f83 DB |
1947 | int __sys_getpeername(int fd, struct sockaddr __user *usockaddr, |
1948 | int __user *usockaddr_len) | |
1da177e4 LT |
1949 | { |
1950 | struct socket *sock; | |
230b1839 | 1951 | struct sockaddr_storage address; |
9b2c45d4 | 1952 | int err, fput_needed; |
1da177e4 | 1953 | |
89bddce5 SH |
1954 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1955 | if (sock != NULL) { | |
1da177e4 LT |
1956 | err = security_socket_getpeername(sock); |
1957 | if (err) { | |
6cb153ca | 1958 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1959 | return err; |
1960 | } | |
1961 | ||
9b2c45d4 DV |
1962 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 1); |
1963 | if (err >= 0) | |
1964 | /* "err" is actually length in this case */ | |
1965 | err = move_addr_to_user(&address, err, usockaddr, | |
89bddce5 | 1966 | usockaddr_len); |
6cb153ca | 1967 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1968 | } |
1969 | return err; | |
1970 | } | |
1971 | ||
b21c8f83 DB |
1972 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1973 | int __user *, usockaddr_len) | |
1974 | { | |
1975 | return __sys_getpeername(fd, usockaddr, usockaddr_len); | |
1976 | } | |
1977 | ||
1da177e4 LT |
1978 | /* |
1979 | * Send a datagram to a given address. We move the address into kernel | |
1980 | * space and check the user space data area is readable before invoking | |
1981 | * the protocol. | |
1982 | */ | |
211b634b DB |
1983 | int __sys_sendto(int fd, void __user *buff, size_t len, unsigned int flags, |
1984 | struct sockaddr __user *addr, int addr_len) | |
1da177e4 LT |
1985 | { |
1986 | struct socket *sock; | |
230b1839 | 1987 | struct sockaddr_storage address; |
1da177e4 LT |
1988 | int err; |
1989 | struct msghdr msg; | |
1990 | struct iovec iov; | |
6cb153ca | 1991 | int fput_needed; |
6cb153ca | 1992 | |
602bd0e9 AV |
1993 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1994 | if (unlikely(err)) | |
1995 | return err; | |
de0fa95c PE |
1996 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1997 | if (!sock) | |
4387ff75 | 1998 | goto out; |
6cb153ca | 1999 | |
89bddce5 | 2000 | msg.msg_name = NULL; |
89bddce5 SH |
2001 | msg.msg_control = NULL; |
2002 | msg.msg_controllen = 0; | |
2003 | msg.msg_namelen = 0; | |
6cb153ca | 2004 | if (addr) { |
43db362d | 2005 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
2006 | if (err < 0) |
2007 | goto out_put; | |
230b1839 | 2008 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 2009 | msg.msg_namelen = addr_len; |
1da177e4 LT |
2010 | } |
2011 | if (sock->file->f_flags & O_NONBLOCK) | |
2012 | flags |= MSG_DONTWAIT; | |
2013 | msg.msg_flags = flags; | |
d8725c86 | 2014 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 2015 | |
89bddce5 | 2016 | out_put: |
de0fa95c | 2017 | fput_light(sock->file, fput_needed); |
4387ff75 | 2018 | out: |
1da177e4 LT |
2019 | return err; |
2020 | } | |
2021 | ||
211b634b DB |
2022 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
2023 | unsigned int, flags, struct sockaddr __user *, addr, | |
2024 | int, addr_len) | |
2025 | { | |
2026 | return __sys_sendto(fd, buff, len, flags, addr, addr_len); | |
2027 | } | |
2028 | ||
1da177e4 | 2029 | /* |
89bddce5 | 2030 | * Send a datagram down a socket. |
1da177e4 LT |
2031 | */ |
2032 | ||
3e0fa65f | 2033 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 2034 | unsigned int, flags) |
1da177e4 | 2035 | { |
211b634b | 2036 | return __sys_sendto(fd, buff, len, flags, NULL, 0); |
1da177e4 LT |
2037 | } |
2038 | ||
2039 | /* | |
89bddce5 | 2040 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
2041 | * sender. We verify the buffers are writable and if needed move the |
2042 | * sender address from kernel to user space. | |
2043 | */ | |
7a09e1eb DB |
2044 | int __sys_recvfrom(int fd, void __user *ubuf, size_t size, unsigned int flags, |
2045 | struct sockaddr __user *addr, int __user *addr_len) | |
1da177e4 LT |
2046 | { |
2047 | struct socket *sock; | |
2048 | struct iovec iov; | |
2049 | struct msghdr msg; | |
230b1839 | 2050 | struct sockaddr_storage address; |
89bddce5 | 2051 | int err, err2; |
6cb153ca BL |
2052 | int fput_needed; |
2053 | ||
602bd0e9 AV |
2054 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
2055 | if (unlikely(err)) | |
2056 | return err; | |
de0fa95c | 2057 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 2058 | if (!sock) |
de0fa95c | 2059 | goto out; |
1da177e4 | 2060 | |
89bddce5 SH |
2061 | msg.msg_control = NULL; |
2062 | msg.msg_controllen = 0; | |
f3d33426 HFS |
2063 | /* Save some cycles and don't copy the address if not needed */ |
2064 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
2065 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
2066 | msg.msg_namelen = 0; | |
130ed5d1 | 2067 | msg.msg_iocb = NULL; |
9f138fa6 | 2068 | msg.msg_flags = 0; |
1da177e4 LT |
2069 | if (sock->file->f_flags & O_NONBLOCK) |
2070 | flags |= MSG_DONTWAIT; | |
2da62906 | 2071 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 2072 | |
89bddce5 | 2073 | if (err >= 0 && addr != NULL) { |
43db362d | 2074 | err2 = move_addr_to_user(&address, |
230b1839 | 2075 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
2076 | if (err2 < 0) |
2077 | err = err2; | |
1da177e4 | 2078 | } |
de0fa95c PE |
2079 | |
2080 | fput_light(sock->file, fput_needed); | |
4387ff75 | 2081 | out: |
1da177e4 LT |
2082 | return err; |
2083 | } | |
2084 | ||
7a09e1eb DB |
2085 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
2086 | unsigned int, flags, struct sockaddr __user *, addr, | |
2087 | int __user *, addr_len) | |
2088 | { | |
2089 | return __sys_recvfrom(fd, ubuf, size, flags, addr, addr_len); | |
2090 | } | |
2091 | ||
1da177e4 | 2092 | /* |
89bddce5 | 2093 | * Receive a datagram from a socket. |
1da177e4 LT |
2094 | */ |
2095 | ||
b7c0ddf5 JG |
2096 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
2097 | unsigned int, flags) | |
1da177e4 | 2098 | { |
7a09e1eb | 2099 | return __sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
1da177e4 LT |
2100 | } |
2101 | ||
83f0c10b FW |
2102 | static bool sock_use_custom_sol_socket(const struct socket *sock) |
2103 | { | |
2104 | const struct sock *sk = sock->sk; | |
2105 | ||
2106 | /* Use sock->ops->setsockopt() for MPTCP */ | |
2107 | return IS_ENABLED(CONFIG_MPTCP) && | |
2108 | sk->sk_protocol == IPPROTO_MPTCP && | |
2109 | sk->sk_type == SOCK_STREAM && | |
2110 | (sk->sk_family == AF_INET || sk->sk_family == AF_INET6); | |
2111 | } | |
2112 | ||
1da177e4 LT |
2113 | /* |
2114 | * Set a socket option. Because we don't know the option lengths we have | |
2115 | * to pass the user mode parameter for the protocols to sort out. | |
2116 | */ | |
a7b75c5a | 2117 | int __sys_setsockopt(int fd, int level, int optname, char __user *user_optval, |
55db9c0e | 2118 | int optlen) |
1da177e4 | 2119 | { |
519a8a6c | 2120 | sockptr_t optval = USER_SOCKPTR(user_optval); |
0d01da6a | 2121 | char *kernel_optval = NULL; |
6cb153ca | 2122 | int err, fput_needed; |
1da177e4 LT |
2123 | struct socket *sock; |
2124 | ||
2125 | if (optlen < 0) | |
2126 | return -EINVAL; | |
89bddce5 SH |
2127 | |
2128 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
4a367299 CH |
2129 | if (!sock) |
2130 | return err; | |
1da177e4 | 2131 | |
4a367299 CH |
2132 | err = security_socket_setsockopt(sock, level, optname); |
2133 | if (err) | |
2134 | goto out_put; | |
0d01da6a | 2135 | |
55db9c0e CH |
2136 | if (!in_compat_syscall()) |
2137 | err = BPF_CGROUP_RUN_PROG_SETSOCKOPT(sock->sk, &level, &optname, | |
a7b75c5a | 2138 | user_optval, &optlen, |
55db9c0e | 2139 | &kernel_optval); |
4a367299 CH |
2140 | if (err < 0) |
2141 | goto out_put; | |
2142 | if (err > 0) { | |
2143 | err = 0; | |
2144 | goto out_put; | |
2145 | } | |
0d01da6a | 2146 | |
a7b75c5a CH |
2147 | if (kernel_optval) |
2148 | optval = KERNEL_SOCKPTR(kernel_optval); | |
4a367299 | 2149 | if (level == SOL_SOCKET && !sock_use_custom_sol_socket(sock)) |
a7b75c5a | 2150 | err = sock_setsockopt(sock, level, optname, optval, optlen); |
a44d9e72 CH |
2151 | else if (unlikely(!sock->ops->setsockopt)) |
2152 | err = -EOPNOTSUPP; | |
4a367299 CH |
2153 | else |
2154 | err = sock->ops->setsockopt(sock, level, optname, optval, | |
89bddce5 | 2155 | optlen); |
a7b75c5a | 2156 | kfree(kernel_optval); |
4a367299 CH |
2157 | out_put: |
2158 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2159 | return err; |
2160 | } | |
2161 | ||
cc36dca0 DB |
2162 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
2163 | char __user *, optval, int, optlen) | |
2164 | { | |
2165 | return __sys_setsockopt(fd, level, optname, optval, optlen); | |
2166 | } | |
2167 | ||
9cacf81f SF |
2168 | INDIRECT_CALLABLE_DECLARE(bool tcp_bpf_bypass_getsockopt(int level, |
2169 | int optname)); | |
2170 | ||
1da177e4 LT |
2171 | /* |
2172 | * Get a socket option. Because we don't know the option lengths we have | |
2173 | * to pass a user mode parameter for the protocols to sort out. | |
2174 | */ | |
55db9c0e CH |
2175 | int __sys_getsockopt(int fd, int level, int optname, char __user *optval, |
2176 | int __user *optlen) | |
1da177e4 | 2177 | { |
6cb153ca | 2178 | int err, fput_needed; |
1da177e4 | 2179 | struct socket *sock; |
0d01da6a | 2180 | int max_optlen; |
1da177e4 | 2181 | |
89bddce5 | 2182 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
d8a9b38f CH |
2183 | if (!sock) |
2184 | return err; | |
2185 | ||
2186 | err = security_socket_getsockopt(sock, level, optname); | |
2187 | if (err) | |
2188 | goto out_put; | |
1da177e4 | 2189 | |
55db9c0e CH |
2190 | if (!in_compat_syscall()) |
2191 | max_optlen = BPF_CGROUP_GETSOCKOPT_MAX_OPTLEN(optlen); | |
0d01da6a | 2192 | |
d8a9b38f CH |
2193 | if (level == SOL_SOCKET) |
2194 | err = sock_getsockopt(sock, level, optname, optval, optlen); | |
a44d9e72 CH |
2195 | else if (unlikely(!sock->ops->getsockopt)) |
2196 | err = -EOPNOTSUPP; | |
d8a9b38f CH |
2197 | else |
2198 | err = sock->ops->getsockopt(sock, level, optname, optval, | |
89bddce5 | 2199 | optlen); |
0d01da6a | 2200 | |
55db9c0e CH |
2201 | if (!in_compat_syscall()) |
2202 | err = BPF_CGROUP_RUN_PROG_GETSOCKOPT(sock->sk, level, optname, | |
2203 | optval, optlen, max_optlen, | |
2204 | err); | |
6cb153ca | 2205 | out_put: |
d8a9b38f | 2206 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2207 | return err; |
2208 | } | |
2209 | ||
13a2d70e DB |
2210 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
2211 | char __user *, optval, int __user *, optlen) | |
2212 | { | |
2213 | return __sys_getsockopt(fd, level, optname, optval, optlen); | |
2214 | } | |
2215 | ||
1da177e4 LT |
2216 | /* |
2217 | * Shutdown a socket. | |
2218 | */ | |
2219 | ||
b713c195 JA |
2220 | int __sys_shutdown_sock(struct socket *sock, int how) |
2221 | { | |
2222 | int err; | |
2223 | ||
2224 | err = security_socket_shutdown(sock, how); | |
2225 | if (!err) | |
2226 | err = sock->ops->shutdown(sock, how); | |
2227 | ||
2228 | return err; | |
2229 | } | |
2230 | ||
005a1aea | 2231 | int __sys_shutdown(int fd, int how) |
1da177e4 | 2232 | { |
6cb153ca | 2233 | int err, fput_needed; |
1da177e4 LT |
2234 | struct socket *sock; |
2235 | ||
89bddce5 SH |
2236 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2237 | if (sock != NULL) { | |
b713c195 | 2238 | err = __sys_shutdown_sock(sock, how); |
6cb153ca | 2239 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2240 | } |
2241 | return err; | |
2242 | } | |
2243 | ||
005a1aea DB |
2244 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
2245 | { | |
2246 | return __sys_shutdown(fd, how); | |
2247 | } | |
2248 | ||
89bddce5 | 2249 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
2250 | * fields which are the same type (int / unsigned) on our platforms. |
2251 | */ | |
2252 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
2253 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
2254 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
2255 | ||
c71d8ebe TH |
2256 | struct used_address { |
2257 | struct sockaddr_storage name; | |
2258 | unsigned int name_len; | |
2259 | }; | |
2260 | ||
0a384abf JA |
2261 | int __copy_msghdr_from_user(struct msghdr *kmsg, |
2262 | struct user_msghdr __user *umsg, | |
2263 | struct sockaddr __user **save_addr, | |
2264 | struct iovec __user **uiov, size_t *nsegs) | |
1661bf36 | 2265 | { |
ffb07550 | 2266 | struct user_msghdr msg; |
08adb7da AV |
2267 | ssize_t err; |
2268 | ||
ffb07550 | 2269 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 2270 | return -EFAULT; |
dbb490b9 | 2271 | |
1f466e1f CH |
2272 | kmsg->msg_control_is_user = true; |
2273 | kmsg->msg_control_user = msg.msg_control; | |
ffb07550 AV |
2274 | kmsg->msg_controllen = msg.msg_controllen; |
2275 | kmsg->msg_flags = msg.msg_flags; | |
2276 | ||
2277 | kmsg->msg_namelen = msg.msg_namelen; | |
2278 | if (!msg.msg_name) | |
6a2a2b3a AS |
2279 | kmsg->msg_namelen = 0; |
2280 | ||
dbb490b9 ML |
2281 | if (kmsg->msg_namelen < 0) |
2282 | return -EINVAL; | |
2283 | ||
1661bf36 | 2284 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 2285 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
2286 | |
2287 | if (save_addr) | |
ffb07550 | 2288 | *save_addr = msg.msg_name; |
08adb7da | 2289 | |
ffb07550 | 2290 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 2291 | if (!save_addr) { |
864d9664 PA |
2292 | err = move_addr_to_kernel(msg.msg_name, |
2293 | kmsg->msg_namelen, | |
08adb7da AV |
2294 | kmsg->msg_name); |
2295 | if (err < 0) | |
2296 | return err; | |
2297 | } | |
2298 | } else { | |
2299 | kmsg->msg_name = NULL; | |
2300 | kmsg->msg_namelen = 0; | |
2301 | } | |
2302 | ||
ffb07550 | 2303 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
2304 | return -EMSGSIZE; |
2305 | ||
0345f931 | 2306 | kmsg->msg_iocb = NULL; |
0a384abf JA |
2307 | *uiov = msg.msg_iov; |
2308 | *nsegs = msg.msg_iovlen; | |
2309 | return 0; | |
2310 | } | |
2311 | ||
2312 | static int copy_msghdr_from_user(struct msghdr *kmsg, | |
2313 | struct user_msghdr __user *umsg, | |
2314 | struct sockaddr __user **save_addr, | |
2315 | struct iovec **iov) | |
2316 | { | |
2317 | struct user_msghdr msg; | |
2318 | ssize_t err; | |
2319 | ||
2320 | err = __copy_msghdr_from_user(kmsg, umsg, save_addr, &msg.msg_iov, | |
2321 | &msg.msg_iovlen); | |
2322 | if (err) | |
2323 | return err; | |
0345f931 | 2324 | |
87e5e6da | 2325 | err = import_iovec(save_addr ? READ : WRITE, |
ffb07550 | 2326 | msg.msg_iov, msg.msg_iovlen, |
da184284 | 2327 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
87e5e6da | 2328 | return err < 0 ? err : 0; |
1661bf36 DC |
2329 | } |
2330 | ||
4257c8ca JA |
2331 | static int ____sys_sendmsg(struct socket *sock, struct msghdr *msg_sys, |
2332 | unsigned int flags, struct used_address *used_address, | |
2333 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 2334 | { |
b9d717a7 | 2335 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 2336 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 2337 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 2338 | unsigned char *ctl_buf = ctl; |
d8725c86 | 2339 | int ctl_len; |
08adb7da | 2340 | ssize_t err; |
89bddce5 | 2341 | |
1da177e4 LT |
2342 | err = -ENOBUFS; |
2343 | ||
228e548e | 2344 | if (msg_sys->msg_controllen > INT_MAX) |
4257c8ca | 2345 | goto out; |
28a94d8f | 2346 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2347 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2348 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2349 | err = |
228e548e | 2350 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2351 | sizeof(ctl)); |
1da177e4 | 2352 | if (err) |
4257c8ca | 2353 | goto out; |
228e548e AB |
2354 | ctl_buf = msg_sys->msg_control; |
2355 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2356 | } else if (ctl_len) { |
ac4340fc DM |
2357 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2358 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2359 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2360 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2361 | if (ctl_buf == NULL) |
4257c8ca | 2362 | goto out; |
1da177e4 LT |
2363 | } |
2364 | err = -EFAULT; | |
1f466e1f | 2365 | if (copy_from_user(ctl_buf, msg_sys->msg_control_user, ctl_len)) |
1da177e4 | 2366 | goto out_freectl; |
228e548e | 2367 | msg_sys->msg_control = ctl_buf; |
1f466e1f | 2368 | msg_sys->msg_control_is_user = false; |
1da177e4 | 2369 | } |
228e548e | 2370 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2371 | |
2372 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2373 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2374 | /* |
2375 | * If this is sendmmsg() and current destination address is same as | |
2376 | * previously succeeded address, omit asking LSM's decision. | |
2377 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2378 | * destination address never matches. | |
2379 | */ | |
bc909d9d MD |
2380 | if (used_address && msg_sys->msg_name && |
2381 | used_address->name_len == msg_sys->msg_namelen && | |
2382 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2383 | used_address->name_len)) { |
d8725c86 | 2384 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2385 | goto out_freectl; |
2386 | } | |
d8725c86 | 2387 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2388 | /* |
2389 | * If this is sendmmsg() and sending to current destination address was | |
2390 | * successful, remember it. | |
2391 | */ | |
2392 | if (used_address && err >= 0) { | |
2393 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2394 | if (msg_sys->msg_name) |
2395 | memcpy(&used_address->name, msg_sys->msg_name, | |
2396 | used_address->name_len); | |
c71d8ebe | 2397 | } |
1da177e4 LT |
2398 | |
2399 | out_freectl: | |
89bddce5 | 2400 | if (ctl_buf != ctl) |
1da177e4 | 2401 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
4257c8ca JA |
2402 | out: |
2403 | return err; | |
2404 | } | |
2405 | ||
03b1230c JA |
2406 | int sendmsg_copy_msghdr(struct msghdr *msg, |
2407 | struct user_msghdr __user *umsg, unsigned flags, | |
2408 | struct iovec **iov) | |
4257c8ca JA |
2409 | { |
2410 | int err; | |
2411 | ||
2412 | if (flags & MSG_CMSG_COMPAT) { | |
2413 | struct compat_msghdr __user *msg_compat; | |
2414 | ||
2415 | msg_compat = (struct compat_msghdr __user *) umsg; | |
2416 | err = get_compat_msghdr(msg, msg_compat, NULL, iov); | |
2417 | } else { | |
2418 | err = copy_msghdr_from_user(msg, umsg, NULL, iov); | |
2419 | } | |
2420 | if (err < 0) | |
2421 | return err; | |
2422 | ||
2423 | return 0; | |
2424 | } | |
2425 | ||
2426 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, | |
2427 | struct msghdr *msg_sys, unsigned int flags, | |
2428 | struct used_address *used_address, | |
2429 | unsigned int allowed_msghdr_flags) | |
2430 | { | |
2431 | struct sockaddr_storage address; | |
2432 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; | |
2433 | ssize_t err; | |
2434 | ||
2435 | msg_sys->msg_name = &address; | |
2436 | ||
2437 | err = sendmsg_copy_msghdr(msg_sys, msg, flags, &iov); | |
2438 | if (err < 0) | |
2439 | return err; | |
2440 | ||
2441 | err = ____sys_sendmsg(sock, msg_sys, flags, used_address, | |
2442 | allowed_msghdr_flags); | |
da184284 | 2443 | kfree(iov); |
228e548e AB |
2444 | return err; |
2445 | } | |
2446 | ||
2447 | /* | |
2448 | * BSD sendmsg interface | |
2449 | */ | |
03b1230c | 2450 | long __sys_sendmsg_sock(struct socket *sock, struct msghdr *msg, |
0fa03c62 JA |
2451 | unsigned int flags) |
2452 | { | |
03b1230c | 2453 | return ____sys_sendmsg(sock, msg, flags, NULL, 0); |
0fa03c62 | 2454 | } |
228e548e | 2455 | |
e1834a32 DB |
2456 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2457 | bool forbid_cmsg_compat) | |
228e548e AB |
2458 | { |
2459 | int fput_needed, err; | |
2460 | struct msghdr msg_sys; | |
1be374a0 AL |
2461 | struct socket *sock; |
2462 | ||
e1834a32 DB |
2463 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2464 | return -EINVAL; | |
2465 | ||
1be374a0 | 2466 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2467 | if (!sock) |
2468 | goto out; | |
2469 | ||
28a94d8f | 2470 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2471 | |
6cb153ca | 2472 | fput_light(sock->file, fput_needed); |
89bddce5 | 2473 | out: |
1da177e4 LT |
2474 | return err; |
2475 | } | |
2476 | ||
666547ff | 2477 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 | 2478 | { |
e1834a32 | 2479 | return __sys_sendmsg(fd, msg, flags, true); |
a7526eb5 AL |
2480 | } |
2481 | ||
228e548e AB |
2482 | /* |
2483 | * Linux sendmmsg interface | |
2484 | */ | |
2485 | ||
2486 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
e1834a32 | 2487 | unsigned int flags, bool forbid_cmsg_compat) |
228e548e AB |
2488 | { |
2489 | int fput_needed, err, datagrams; | |
2490 | struct socket *sock; | |
2491 | struct mmsghdr __user *entry; | |
2492 | struct compat_mmsghdr __user *compat_entry; | |
2493 | struct msghdr msg_sys; | |
c71d8ebe | 2494 | struct used_address used_address; |
f092276d | 2495 | unsigned int oflags = flags; |
228e548e | 2496 | |
e1834a32 DB |
2497 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2498 | return -EINVAL; | |
2499 | ||
98382f41 AB |
2500 | if (vlen > UIO_MAXIOV) |
2501 | vlen = UIO_MAXIOV; | |
228e548e AB |
2502 | |
2503 | datagrams = 0; | |
2504 | ||
2505 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2506 | if (!sock) | |
2507 | return err; | |
2508 | ||
c71d8ebe | 2509 | used_address.name_len = UINT_MAX; |
228e548e AB |
2510 | entry = mmsg; |
2511 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2512 | err = 0; |
f092276d | 2513 | flags |= MSG_BATCH; |
228e548e AB |
2514 | |
2515 | while (datagrams < vlen) { | |
f092276d TH |
2516 | if (datagrams == vlen - 1) |
2517 | flags = oflags; | |
2518 | ||
228e548e | 2519 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2520 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2521 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2522 | if (err < 0) |
2523 | break; | |
2524 | err = __put_user(err, &compat_entry->msg_len); | |
2525 | ++compat_entry; | |
2526 | } else { | |
a7526eb5 | 2527 | err = ___sys_sendmsg(sock, |
666547ff | 2528 | (struct user_msghdr __user *)entry, |
28a94d8f | 2529 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2530 | if (err < 0) |
2531 | break; | |
2532 | err = put_user(err, &entry->msg_len); | |
2533 | ++entry; | |
2534 | } | |
2535 | ||
2536 | if (err) | |
2537 | break; | |
2538 | ++datagrams; | |
3023898b SHY |
2539 | if (msg_data_left(&msg_sys)) |
2540 | break; | |
a78cb84c | 2541 | cond_resched(); |
228e548e AB |
2542 | } |
2543 | ||
228e548e AB |
2544 | fput_light(sock->file, fput_needed); |
2545 | ||
728ffb86 AB |
2546 | /* We only return an error if no datagrams were able to be sent */ |
2547 | if (datagrams != 0) | |
228e548e AB |
2548 | return datagrams; |
2549 | ||
228e548e AB |
2550 | return err; |
2551 | } | |
2552 | ||
2553 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2554 | unsigned int, vlen, unsigned int, flags) | |
2555 | { | |
e1834a32 | 2556 | return __sys_sendmmsg(fd, mmsg, vlen, flags, true); |
228e548e AB |
2557 | } |
2558 | ||
03b1230c JA |
2559 | int recvmsg_copy_msghdr(struct msghdr *msg, |
2560 | struct user_msghdr __user *umsg, unsigned flags, | |
2561 | struct sockaddr __user **uaddr, | |
2562 | struct iovec **iov) | |
1da177e4 | 2563 | { |
08adb7da | 2564 | ssize_t err; |
1da177e4 | 2565 | |
4257c8ca JA |
2566 | if (MSG_CMSG_COMPAT & flags) { |
2567 | struct compat_msghdr __user *msg_compat; | |
1da177e4 | 2568 | |
4257c8ca JA |
2569 | msg_compat = (struct compat_msghdr __user *) umsg; |
2570 | err = get_compat_msghdr(msg, msg_compat, uaddr, iov); | |
2571 | } else { | |
2572 | err = copy_msghdr_from_user(msg, umsg, uaddr, iov); | |
2573 | } | |
1da177e4 | 2574 | if (err < 0) |
da184284 | 2575 | return err; |
1da177e4 | 2576 | |
4257c8ca JA |
2577 | return 0; |
2578 | } | |
2579 | ||
2580 | static int ____sys_recvmsg(struct socket *sock, struct msghdr *msg_sys, | |
2581 | struct user_msghdr __user *msg, | |
2582 | struct sockaddr __user *uaddr, | |
2583 | unsigned int flags, int nosec) | |
2584 | { | |
2585 | struct compat_msghdr __user *msg_compat = | |
2586 | (struct compat_msghdr __user *) msg; | |
2587 | int __user *uaddr_len = COMPAT_NAMELEN(msg); | |
2588 | struct sockaddr_storage addr; | |
2589 | unsigned long cmsg_ptr; | |
2590 | int len; | |
2591 | ssize_t err; | |
2592 | ||
2593 | msg_sys->msg_name = &addr; | |
a2e27255 ACM |
2594 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2595 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2596 | |
f3d33426 HFS |
2597 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2598 | msg_sys->msg_namelen = 0; | |
2599 | ||
1da177e4 LT |
2600 | if (sock->file->f_flags & O_NONBLOCK) |
2601 | flags |= MSG_DONTWAIT; | |
1af66221 ED |
2602 | |
2603 | if (unlikely(nosec)) | |
2604 | err = sock_recvmsg_nosec(sock, msg_sys, flags); | |
2605 | else | |
2606 | err = sock_recvmsg(sock, msg_sys, flags); | |
2607 | ||
1da177e4 | 2608 | if (err < 0) |
4257c8ca | 2609 | goto out; |
1da177e4 LT |
2610 | len = err; |
2611 | ||
2612 | if (uaddr != NULL) { | |
43db362d | 2613 | err = move_addr_to_user(&addr, |
a2e27255 | 2614 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2615 | uaddr_len); |
1da177e4 | 2616 | if (err < 0) |
4257c8ca | 2617 | goto out; |
1da177e4 | 2618 | } |
a2e27255 | 2619 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2620 | COMPAT_FLAGS(msg)); |
1da177e4 | 2621 | if (err) |
4257c8ca | 2622 | goto out; |
1da177e4 | 2623 | if (MSG_CMSG_COMPAT & flags) |
a2e27255 | 2624 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2625 | &msg_compat->msg_controllen); |
2626 | else | |
a2e27255 | 2627 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2628 | &msg->msg_controllen); |
2629 | if (err) | |
4257c8ca | 2630 | goto out; |
1da177e4 | 2631 | err = len; |
4257c8ca JA |
2632 | out: |
2633 | return err; | |
2634 | } | |
2635 | ||
2636 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, | |
2637 | struct msghdr *msg_sys, unsigned int flags, int nosec) | |
2638 | { | |
2639 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; | |
2640 | /* user mode address pointers */ | |
2641 | struct sockaddr __user *uaddr; | |
2642 | ssize_t err; | |
2643 | ||
2644 | err = recvmsg_copy_msghdr(msg_sys, msg, flags, &uaddr, &iov); | |
2645 | if (err < 0) | |
2646 | return err; | |
1da177e4 | 2647 | |
4257c8ca | 2648 | err = ____sys_recvmsg(sock, msg_sys, msg, uaddr, flags, nosec); |
da184284 | 2649 | kfree(iov); |
a2e27255 ACM |
2650 | return err; |
2651 | } | |
2652 | ||
2653 | /* | |
2654 | * BSD recvmsg interface | |
2655 | */ | |
2656 | ||
03b1230c JA |
2657 | long __sys_recvmsg_sock(struct socket *sock, struct msghdr *msg, |
2658 | struct user_msghdr __user *umsg, | |
2659 | struct sockaddr __user *uaddr, unsigned int flags) | |
aa1fa28f | 2660 | { |
03b1230c | 2661 | return ____sys_recvmsg(sock, msg, umsg, uaddr, flags, 0); |
aa1fa28f JA |
2662 | } |
2663 | ||
e1834a32 DB |
2664 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2665 | bool forbid_cmsg_compat) | |
a2e27255 ACM |
2666 | { |
2667 | int fput_needed, err; | |
2668 | struct msghdr msg_sys; | |
1be374a0 AL |
2669 | struct socket *sock; |
2670 | ||
e1834a32 DB |
2671 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2672 | return -EINVAL; | |
2673 | ||
1be374a0 | 2674 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2675 | if (!sock) |
2676 | goto out; | |
2677 | ||
a7526eb5 | 2678 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2679 | |
6cb153ca | 2680 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2681 | out: |
2682 | return err; | |
2683 | } | |
2684 | ||
666547ff | 2685 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2686 | unsigned int, flags) |
2687 | { | |
e1834a32 | 2688 | return __sys_recvmsg(fd, msg, flags, true); |
a7526eb5 AL |
2689 | } |
2690 | ||
a2e27255 ACM |
2691 | /* |
2692 | * Linux recvmmsg interface | |
2693 | */ | |
2694 | ||
e11d4284 AB |
2695 | static int do_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2696 | unsigned int vlen, unsigned int flags, | |
2697 | struct timespec64 *timeout) | |
a2e27255 ACM |
2698 | { |
2699 | int fput_needed, err, datagrams; | |
2700 | struct socket *sock; | |
2701 | struct mmsghdr __user *entry; | |
d7256d0e | 2702 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2703 | struct msghdr msg_sys; |
766b9f92 DD |
2704 | struct timespec64 end_time; |
2705 | struct timespec64 timeout64; | |
a2e27255 ACM |
2706 | |
2707 | if (timeout && | |
2708 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2709 | timeout->tv_nsec)) | |
2710 | return -EINVAL; | |
2711 | ||
2712 | datagrams = 0; | |
2713 | ||
2714 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2715 | if (!sock) | |
2716 | return err; | |
2717 | ||
7797dc41 SHY |
2718 | if (likely(!(flags & MSG_ERRQUEUE))) { |
2719 | err = sock_error(sock->sk); | |
2720 | if (err) { | |
2721 | datagrams = err; | |
2722 | goto out_put; | |
2723 | } | |
e623a9e9 | 2724 | } |
a2e27255 ACM |
2725 | |
2726 | entry = mmsg; | |
d7256d0e | 2727 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2728 | |
2729 | while (datagrams < vlen) { | |
2730 | /* | |
2731 | * No need to ask LSM for more than the first datagram. | |
2732 | */ | |
d7256d0e | 2733 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2734 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2735 | &msg_sys, flags & ~MSG_WAITFORONE, |
2736 | datagrams); | |
d7256d0e JMG |
2737 | if (err < 0) |
2738 | break; | |
2739 | err = __put_user(err, &compat_entry->msg_len); | |
2740 | ++compat_entry; | |
2741 | } else { | |
a7526eb5 | 2742 | err = ___sys_recvmsg(sock, |
666547ff | 2743 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2744 | &msg_sys, flags & ~MSG_WAITFORONE, |
2745 | datagrams); | |
d7256d0e JMG |
2746 | if (err < 0) |
2747 | break; | |
2748 | err = put_user(err, &entry->msg_len); | |
2749 | ++entry; | |
2750 | } | |
2751 | ||
a2e27255 ACM |
2752 | if (err) |
2753 | break; | |
a2e27255 ACM |
2754 | ++datagrams; |
2755 | ||
71c5c159 BB |
2756 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2757 | if (flags & MSG_WAITFORONE) | |
2758 | flags |= MSG_DONTWAIT; | |
2759 | ||
a2e27255 | 2760 | if (timeout) { |
766b9f92 | 2761 | ktime_get_ts64(&timeout64); |
c2e6c856 | 2762 | *timeout = timespec64_sub(end_time, timeout64); |
a2e27255 ACM |
2763 | if (timeout->tv_sec < 0) { |
2764 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2765 | break; | |
2766 | } | |
2767 | ||
2768 | /* Timeout, return less than vlen datagrams */ | |
2769 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2770 | break; | |
2771 | } | |
2772 | ||
2773 | /* Out of band data, return right away */ | |
2774 | if (msg_sys.msg_flags & MSG_OOB) | |
2775 | break; | |
a78cb84c | 2776 | cond_resched(); |
a2e27255 ACM |
2777 | } |
2778 | ||
a2e27255 | 2779 | if (err == 0) |
34b88a68 ACM |
2780 | goto out_put; |
2781 | ||
2782 | if (datagrams == 0) { | |
2783 | datagrams = err; | |
2784 | goto out_put; | |
2785 | } | |
a2e27255 | 2786 | |
34b88a68 ACM |
2787 | /* |
2788 | * We may return less entries than requested (vlen) if the | |
2789 | * sock is non block and there aren't enough datagrams... | |
2790 | */ | |
2791 | if (err != -EAGAIN) { | |
a2e27255 | 2792 | /* |
34b88a68 ACM |
2793 | * ... or if recvmsg returns an error after we |
2794 | * received some datagrams, where we record the | |
2795 | * error to return on the next call or if the | |
2796 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2797 | */ |
34b88a68 | 2798 | sock->sk->sk_err = -err; |
a2e27255 | 2799 | } |
34b88a68 ACM |
2800 | out_put: |
2801 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2802 | |
34b88a68 | 2803 | return datagrams; |
a2e27255 ACM |
2804 | } |
2805 | ||
e11d4284 AB |
2806 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2807 | unsigned int vlen, unsigned int flags, | |
2808 | struct __kernel_timespec __user *timeout, | |
2809 | struct old_timespec32 __user *timeout32) | |
a2e27255 ACM |
2810 | { |
2811 | int datagrams; | |
c2e6c856 | 2812 | struct timespec64 timeout_sys; |
a2e27255 | 2813 | |
e11d4284 AB |
2814 | if (timeout && get_timespec64(&timeout_sys, timeout)) |
2815 | return -EFAULT; | |
a2e27255 | 2816 | |
e11d4284 | 2817 | if (timeout32 && get_old_timespec32(&timeout_sys, timeout32)) |
a2e27255 ACM |
2818 | return -EFAULT; |
2819 | ||
e11d4284 AB |
2820 | if (!timeout && !timeout32) |
2821 | return do_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2822 | ||
2823 | datagrams = do_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
a2e27255 | 2824 | |
e11d4284 AB |
2825 | if (datagrams <= 0) |
2826 | return datagrams; | |
2827 | ||
2828 | if (timeout && put_timespec64(&timeout_sys, timeout)) | |
2829 | datagrams = -EFAULT; | |
2830 | ||
2831 | if (timeout32 && put_old_timespec32(&timeout_sys, timeout32)) | |
a2e27255 ACM |
2832 | datagrams = -EFAULT; |
2833 | ||
2834 | return datagrams; | |
2835 | } | |
2836 | ||
1255e269 DB |
2837 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, |
2838 | unsigned int, vlen, unsigned int, flags, | |
c2e6c856 | 2839 | struct __kernel_timespec __user *, timeout) |
1255e269 | 2840 | { |
e11d4284 AB |
2841 | if (flags & MSG_CMSG_COMPAT) |
2842 | return -EINVAL; | |
2843 | ||
2844 | return __sys_recvmmsg(fd, mmsg, vlen, flags, timeout, NULL); | |
2845 | } | |
2846 | ||
2847 | #ifdef CONFIG_COMPAT_32BIT_TIME | |
2848 | SYSCALL_DEFINE5(recvmmsg_time32, int, fd, struct mmsghdr __user *, mmsg, | |
2849 | unsigned int, vlen, unsigned int, flags, | |
2850 | struct old_timespec32 __user *, timeout) | |
2851 | { | |
2852 | if (flags & MSG_CMSG_COMPAT) | |
2853 | return -EINVAL; | |
2854 | ||
2855 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL, timeout); | |
1255e269 | 2856 | } |
e11d4284 | 2857 | #endif |
1255e269 | 2858 | |
a2e27255 | 2859 | #ifdef __ARCH_WANT_SYS_SOCKETCALL |
1da177e4 LT |
2860 | /* Argument list sizes for sys_socketcall */ |
2861 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2862 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2863 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2864 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2865 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2866 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2867 | }; |
2868 | ||
1da177e4 LT |
2869 | #undef AL |
2870 | ||
2871 | /* | |
89bddce5 | 2872 | * System call vectors. |
1da177e4 LT |
2873 | * |
2874 | * Argument checking cleaned up. Saved 20% in size. | |
2875 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2876 | * it is set by the callees. |
1da177e4 LT |
2877 | */ |
2878 | ||
3e0fa65f | 2879 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2880 | { |
2950fa9d | 2881 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2882 | unsigned long a0, a1; |
1da177e4 | 2883 | int err; |
47379052 | 2884 | unsigned int len; |
1da177e4 | 2885 | |
228e548e | 2886 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 | 2887 | return -EINVAL; |
c8e8cd57 | 2888 | call = array_index_nospec(call, SYS_SENDMMSG + 1); |
1da177e4 | 2889 | |
47379052 AV |
2890 | len = nargs[call]; |
2891 | if (len > sizeof(a)) | |
2892 | return -EINVAL; | |
2893 | ||
1da177e4 | 2894 | /* copy_from_user should be SMP safe. */ |
47379052 | 2895 | if (copy_from_user(a, args, len)) |
1da177e4 | 2896 | return -EFAULT; |
3ec3b2fb | 2897 | |
2950fa9d CG |
2898 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2899 | if (err) | |
2900 | return err; | |
3ec3b2fb | 2901 | |
89bddce5 SH |
2902 | a0 = a[0]; |
2903 | a1 = a[1]; | |
2904 | ||
2905 | switch (call) { | |
2906 | case SYS_SOCKET: | |
9d6a15c3 | 2907 | err = __sys_socket(a0, a1, a[2]); |
89bddce5 SH |
2908 | break; |
2909 | case SYS_BIND: | |
a87d35d8 | 2910 | err = __sys_bind(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2911 | break; |
2912 | case SYS_CONNECT: | |
1387c2c2 | 2913 | err = __sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
2914 | break; |
2915 | case SYS_LISTEN: | |
25e290ee | 2916 | err = __sys_listen(a0, a1); |
89bddce5 SH |
2917 | break; |
2918 | case SYS_ACCEPT: | |
4541e805 DB |
2919 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2920 | (int __user *)a[2], 0); | |
89bddce5 SH |
2921 | break; |
2922 | case SYS_GETSOCKNAME: | |
2923 | err = | |
8882a107 DB |
2924 | __sys_getsockname(a0, (struct sockaddr __user *)a1, |
2925 | (int __user *)a[2]); | |
89bddce5 SH |
2926 | break; |
2927 | case SYS_GETPEERNAME: | |
2928 | err = | |
b21c8f83 DB |
2929 | __sys_getpeername(a0, (struct sockaddr __user *)a1, |
2930 | (int __user *)a[2]); | |
89bddce5 SH |
2931 | break; |
2932 | case SYS_SOCKETPAIR: | |
6debc8d8 | 2933 | err = __sys_socketpair(a0, a1, a[2], (int __user *)a[3]); |
89bddce5 SH |
2934 | break; |
2935 | case SYS_SEND: | |
f3bf896b DB |
2936 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2937 | NULL, 0); | |
89bddce5 SH |
2938 | break; |
2939 | case SYS_SENDTO: | |
211b634b DB |
2940 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
2941 | (struct sockaddr __user *)a[4], a[5]); | |
89bddce5 SH |
2942 | break; |
2943 | case SYS_RECV: | |
d27e9afc DB |
2944 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2945 | NULL, NULL); | |
89bddce5 SH |
2946 | break; |
2947 | case SYS_RECVFROM: | |
7a09e1eb DB |
2948 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
2949 | (struct sockaddr __user *)a[4], | |
2950 | (int __user *)a[5]); | |
89bddce5 SH |
2951 | break; |
2952 | case SYS_SHUTDOWN: | |
005a1aea | 2953 | err = __sys_shutdown(a0, a1); |
89bddce5 SH |
2954 | break; |
2955 | case SYS_SETSOCKOPT: | |
cc36dca0 DB |
2956 | err = __sys_setsockopt(a0, a1, a[2], (char __user *)a[3], |
2957 | a[4]); | |
89bddce5 SH |
2958 | break; |
2959 | case SYS_GETSOCKOPT: | |
2960 | err = | |
13a2d70e DB |
2961 | __sys_getsockopt(a0, a1, a[2], (char __user *)a[3], |
2962 | (int __user *)a[4]); | |
89bddce5 SH |
2963 | break; |
2964 | case SYS_SENDMSG: | |
e1834a32 DB |
2965 | err = __sys_sendmsg(a0, (struct user_msghdr __user *)a1, |
2966 | a[2], true); | |
89bddce5 | 2967 | break; |
228e548e | 2968 | case SYS_SENDMMSG: |
e1834a32 DB |
2969 | err = __sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
2970 | a[3], true); | |
228e548e | 2971 | break; |
89bddce5 | 2972 | case SYS_RECVMSG: |
e1834a32 DB |
2973 | err = __sys_recvmsg(a0, (struct user_msghdr __user *)a1, |
2974 | a[2], true); | |
89bddce5 | 2975 | break; |
a2e27255 | 2976 | case SYS_RECVMMSG: |
3ca47e95 | 2977 | if (IS_ENABLED(CONFIG_64BIT)) |
e11d4284 AB |
2978 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, |
2979 | a[2], a[3], | |
2980 | (struct __kernel_timespec __user *)a[4], | |
2981 | NULL); | |
2982 | else | |
2983 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, | |
2984 | a[2], a[3], NULL, | |
2985 | (struct old_timespec32 __user *)a[4]); | |
a2e27255 | 2986 | break; |
de11defe | 2987 | case SYS_ACCEPT4: |
4541e805 DB |
2988 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
2989 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2990 | break; |
89bddce5 SH |
2991 | default: |
2992 | err = -EINVAL; | |
2993 | break; | |
1da177e4 LT |
2994 | } |
2995 | return err; | |
2996 | } | |
2997 | ||
89bddce5 | 2998 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2999 | |
55737fda SH |
3000 | /** |
3001 | * sock_register - add a socket protocol handler | |
3002 | * @ops: description of protocol | |
3003 | * | |
1da177e4 LT |
3004 | * This function is called by a protocol handler that wants to |
3005 | * advertise its address family, and have it linked into the | |
e793c0f7 | 3006 | * socket interface. The value ops->family corresponds to the |
55737fda | 3007 | * socket system call protocol family. |
1da177e4 | 3008 | */ |
f0fd27d4 | 3009 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
3010 | { |
3011 | int err; | |
3012 | ||
3013 | if (ops->family >= NPROTO) { | |
3410f22e | 3014 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
3015 | return -ENOBUFS; |
3016 | } | |
55737fda SH |
3017 | |
3018 | spin_lock(&net_family_lock); | |
190683a9 ED |
3019 | if (rcu_dereference_protected(net_families[ops->family], |
3020 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
3021 | err = -EEXIST; |
3022 | else { | |
cf778b00 | 3023 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
3024 | err = 0; |
3025 | } | |
55737fda SH |
3026 | spin_unlock(&net_family_lock); |
3027 | ||
fe0bdbde | 3028 | pr_info("NET: Registered %s protocol family\n", pf_family_names[ops->family]); |
1da177e4 LT |
3029 | return err; |
3030 | } | |
c6d409cf | 3031 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 3032 | |
55737fda SH |
3033 | /** |
3034 | * sock_unregister - remove a protocol handler | |
3035 | * @family: protocol family to remove | |
3036 | * | |
1da177e4 LT |
3037 | * This function is called by a protocol handler that wants to |
3038 | * remove its address family, and have it unlinked from the | |
55737fda SH |
3039 | * new socket creation. |
3040 | * | |
3041 | * If protocol handler is a module, then it can use module reference | |
3042 | * counts to protect against new references. If protocol handler is not | |
3043 | * a module then it needs to provide its own protection in | |
3044 | * the ops->create routine. | |
1da177e4 | 3045 | */ |
f0fd27d4 | 3046 | void sock_unregister(int family) |
1da177e4 | 3047 | { |
f0fd27d4 | 3048 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 3049 | |
55737fda | 3050 | spin_lock(&net_family_lock); |
a9b3cd7f | 3051 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
3052 | spin_unlock(&net_family_lock); |
3053 | ||
3054 | synchronize_rcu(); | |
3055 | ||
fe0bdbde | 3056 | pr_info("NET: Unregistered %s protocol family\n", pf_family_names[family]); |
1da177e4 | 3057 | } |
c6d409cf | 3058 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 3059 | |
bf2ae2e4 XL |
3060 | bool sock_is_registered(int family) |
3061 | { | |
66b51b0a | 3062 | return family < NPROTO && rcu_access_pointer(net_families[family]); |
bf2ae2e4 XL |
3063 | } |
3064 | ||
77d76ea3 | 3065 | static int __init sock_init(void) |
1da177e4 | 3066 | { |
b3e19d92 | 3067 | int err; |
2ca794e5 EB |
3068 | /* |
3069 | * Initialize the network sysctl infrastructure. | |
3070 | */ | |
3071 | err = net_sysctl_init(); | |
3072 | if (err) | |
3073 | goto out; | |
b3e19d92 | 3074 | |
1da177e4 | 3075 | /* |
89bddce5 | 3076 | * Initialize skbuff SLAB cache |
1da177e4 LT |
3077 | */ |
3078 | skb_init(); | |
1da177e4 LT |
3079 | |
3080 | /* | |
89bddce5 | 3081 | * Initialize the protocols module. |
1da177e4 LT |
3082 | */ |
3083 | ||
3084 | init_inodecache(); | |
b3e19d92 NP |
3085 | |
3086 | err = register_filesystem(&sock_fs_type); | |
3087 | if (err) | |
47260ba9 | 3088 | goto out; |
1da177e4 | 3089 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
3090 | if (IS_ERR(sock_mnt)) { |
3091 | err = PTR_ERR(sock_mnt); | |
3092 | goto out_mount; | |
3093 | } | |
77d76ea3 AK |
3094 | |
3095 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
3096 | */ |
3097 | ||
3098 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
3099 | err = netfilter_init(); |
3100 | if (err) | |
3101 | goto out; | |
1da177e4 | 3102 | #endif |
cbeb321a | 3103 | |
408eccce | 3104 | ptp_classifier_init(); |
c1f19b51 | 3105 | |
b3e19d92 NP |
3106 | out: |
3107 | return err; | |
3108 | ||
3109 | out_mount: | |
3110 | unregister_filesystem(&sock_fs_type); | |
b3e19d92 | 3111 | goto out; |
1da177e4 LT |
3112 | } |
3113 | ||
77d76ea3 AK |
3114 | core_initcall(sock_init); /* early initcall */ |
3115 | ||
1da177e4 LT |
3116 | #ifdef CONFIG_PROC_FS |
3117 | void socket_seq_show(struct seq_file *seq) | |
3118 | { | |
648845ab TZ |
3119 | seq_printf(seq, "sockets: used %d\n", |
3120 | sock_inuse_get(seq->private)); | |
1da177e4 | 3121 | } |
89bddce5 | 3122 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 3123 | |
89bbfc95 | 3124 | #ifdef CONFIG_COMPAT |
7a50a240 AB |
3125 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
3126 | { | |
7a50a240 | 3127 | compat_uptr_t uptr32; |
44c02a2c AV |
3128 | struct ifreq ifr; |
3129 | void __user *saved; | |
3130 | int err; | |
7a50a240 | 3131 | |
44c02a2c | 3132 | if (copy_from_user(&ifr, uifr32, sizeof(struct compat_ifreq))) |
7a50a240 AB |
3133 | return -EFAULT; |
3134 | ||
3135 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
3136 | return -EFAULT; | |
3137 | ||
44c02a2c AV |
3138 | saved = ifr.ifr_settings.ifs_ifsu.raw_hdlc; |
3139 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = compat_ptr(uptr32); | |
7a229387 | 3140 | |
44c02a2c AV |
3141 | err = dev_ioctl(net, SIOCWANDEV, &ifr, NULL); |
3142 | if (!err) { | |
3143 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = saved; | |
3144 | if (copy_to_user(uifr32, &ifr, sizeof(struct compat_ifreq))) | |
3145 | err = -EFAULT; | |
ccbd6a5a | 3146 | } |
44c02a2c | 3147 | return err; |
7a229387 AB |
3148 | } |
3149 | ||
590d4693 BH |
3150 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
3151 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 3152 | struct compat_ifreq __user *u_ifreq32) |
7a229387 | 3153 | { |
44c02a2c | 3154 | struct ifreq ifreq; |
7a229387 AB |
3155 | u32 data32; |
3156 | ||
44c02a2c | 3157 | if (copy_from_user(ifreq.ifr_name, u_ifreq32->ifr_name, IFNAMSIZ)) |
7a229387 | 3158 | return -EFAULT; |
44c02a2c | 3159 | if (get_user(data32, &u_ifreq32->ifr_data)) |
7a229387 | 3160 | return -EFAULT; |
44c02a2c | 3161 | ifreq.ifr_data = compat_ptr(data32); |
7a229387 | 3162 | |
44c02a2c | 3163 | return dev_ioctl(net, cmd, &ifreq, NULL); |
7a229387 AB |
3164 | } |
3165 | ||
37ac39bd JB |
3166 | static int compat_ifreq_ioctl(struct net *net, struct socket *sock, |
3167 | unsigned int cmd, | |
3168 | struct compat_ifreq __user *uifr32) | |
3169 | { | |
3170 | struct ifreq __user *uifr; | |
3171 | int err; | |
3172 | ||
3173 | /* Handle the fact that while struct ifreq has the same *layout* on | |
3174 | * 32/64 for everything but ifreq::ifru_ifmap and ifreq::ifru_data, | |
3175 | * which are handled elsewhere, it still has different *size* due to | |
3176 | * ifreq::ifru_ifmap (which is 16 bytes on 32 bit, 24 bytes on 64-bit, | |
3177 | * resulting in struct ifreq being 32 and 40 bytes respectively). | |
3178 | * As a result, if the struct happens to be at the end of a page and | |
3179 | * the next page isn't readable/writable, we get a fault. To prevent | |
3180 | * that, copy back and forth to the full size. | |
3181 | */ | |
3182 | ||
3183 | uifr = compat_alloc_user_space(sizeof(*uifr)); | |
3184 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
3185 | return -EFAULT; | |
3186 | ||
3187 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
3188 | ||
3189 | if (!err) { | |
3190 | switch (cmd) { | |
3191 | case SIOCGIFFLAGS: | |
3192 | case SIOCGIFMETRIC: | |
3193 | case SIOCGIFMTU: | |
3194 | case SIOCGIFMEM: | |
3195 | case SIOCGIFHWADDR: | |
3196 | case SIOCGIFINDEX: | |
3197 | case SIOCGIFADDR: | |
3198 | case SIOCGIFBRDADDR: | |
3199 | case SIOCGIFDSTADDR: | |
3200 | case SIOCGIFNETMASK: | |
3201 | case SIOCGIFPFLAGS: | |
3202 | case SIOCGIFTXQLEN: | |
3203 | case SIOCGMIIPHY: | |
3204 | case SIOCGMIIREG: | |
c6c9fee3 | 3205 | case SIOCGIFNAME: |
37ac39bd JB |
3206 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
3207 | err = -EFAULT; | |
3208 | break; | |
3209 | } | |
3210 | } | |
3211 | return err; | |
3212 | } | |
3213 | ||
7a229387 AB |
3214 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE |
3215 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3216 | * use compatible ioctls |
7a229387 | 3217 | */ |
6b96018b | 3218 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3219 | { |
6b96018b | 3220 | compat_ulong_t tmp; |
7a229387 | 3221 | |
6b96018b | 3222 | if (get_user(tmp, argp)) |
7a229387 AB |
3223 | return -EFAULT; |
3224 | if (tmp == BRCTL_GET_VERSION) | |
3225 | return BRCTL_VERSION + 1; | |
3226 | return -EINVAL; | |
3227 | } | |
3228 | ||
6b96018b AB |
3229 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3230 | unsigned int cmd, unsigned long arg) | |
3231 | { | |
3232 | void __user *argp = compat_ptr(arg); | |
3233 | struct sock *sk = sock->sk; | |
3234 | struct net *net = sock_net(sk); | |
7a229387 | 3235 | |
6b96018b | 3236 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3237 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3238 | |
3239 | switch (cmd) { | |
3240 | case SIOCSIFBR: | |
3241 | case SIOCGIFBR: | |
3242 | return old_bridge_ioctl(argp); | |
7a50a240 AB |
3243 | case SIOCWANDEV: |
3244 | return compat_siocwandev(net, argp); | |
0768e170 AB |
3245 | case SIOCGSTAMP_OLD: |
3246 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
3247 | if (!sock->ops->gettstamp) |
3248 | return -ENOIOCTLCMD; | |
0768e170 | 3249 | return sock->ops->gettstamp(sock, argp, cmd == SIOCGSTAMP_OLD, |
c7cbdbf2 AB |
3250 | !COMPAT_USE_64BIT_TIME); |
3251 | ||
dd98d289 | 3252 | case SIOCETHTOOL: |
590d4693 BH |
3253 | case SIOCBONDSLAVEINFOQUERY: |
3254 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3255 | case SIOCSHWTSTAMP: |
fd468c74 | 3256 | case SIOCGHWTSTAMP: |
590d4693 | 3257 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3258 | |
3259 | case FIOSETOWN: | |
3260 | case SIOCSPGRP: | |
3261 | case FIOGETOWN: | |
3262 | case SIOCGPGRP: | |
3263 | case SIOCBRADDBR: | |
3264 | case SIOCBRDELBR: | |
3265 | case SIOCGIFVLAN: | |
3266 | case SIOCSIFVLAN: | |
c62cce2c | 3267 | case SIOCGSKNS: |
0768e170 AB |
3268 | case SIOCGSTAMP_NEW: |
3269 | case SIOCGSTAMPNS_NEW: | |
876f0bf9 | 3270 | case SIOCGIFCONF: |
6b96018b AB |
3271 | return sock_ioctl(file, cmd, arg); |
3272 | ||
3273 | case SIOCGIFFLAGS: | |
3274 | case SIOCSIFFLAGS: | |
709566d7 AB |
3275 | case SIOCGIFMAP: |
3276 | case SIOCSIFMAP: | |
6b96018b AB |
3277 | case SIOCGIFMETRIC: |
3278 | case SIOCSIFMETRIC: | |
3279 | case SIOCGIFMTU: | |
3280 | case SIOCSIFMTU: | |
3281 | case SIOCGIFMEM: | |
3282 | case SIOCSIFMEM: | |
3283 | case SIOCGIFHWADDR: | |
3284 | case SIOCSIFHWADDR: | |
3285 | case SIOCADDMULTI: | |
3286 | case SIOCDELMULTI: | |
3287 | case SIOCGIFINDEX: | |
6b96018b AB |
3288 | case SIOCGIFADDR: |
3289 | case SIOCSIFADDR: | |
3290 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3291 | case SIOCDIFADDR: |
6b96018b AB |
3292 | case SIOCGIFBRDADDR: |
3293 | case SIOCSIFBRDADDR: | |
3294 | case SIOCGIFDSTADDR: | |
3295 | case SIOCSIFDSTADDR: | |
3296 | case SIOCGIFNETMASK: | |
3297 | case SIOCSIFNETMASK: | |
3298 | case SIOCSIFPFLAGS: | |
3299 | case SIOCGIFPFLAGS: | |
3300 | case SIOCGIFTXQLEN: | |
3301 | case SIOCSIFTXQLEN: | |
3302 | case SIOCBRADDIF: | |
3303 | case SIOCBRDELIF: | |
c6c9fee3 | 3304 | case SIOCGIFNAME: |
9177efd3 AB |
3305 | case SIOCSIFNAME: |
3306 | case SIOCGMIIPHY: | |
3307 | case SIOCGMIIREG: | |
3308 | case SIOCSMIIREG: | |
f92d4fc9 AV |
3309 | case SIOCBONDENSLAVE: |
3310 | case SIOCBONDRELEASE: | |
3311 | case SIOCBONDSETHWADDR: | |
3312 | case SIOCBONDCHANGEACTIVE: | |
37ac39bd JB |
3313 | return compat_ifreq_ioctl(net, sock, cmd, argp); |
3314 | ||
6b96018b AB |
3315 | case SIOCSARP: |
3316 | case SIOCGARP: | |
3317 | case SIOCDARP: | |
c7dc504e | 3318 | case SIOCOUTQ: |
9d7bf41f | 3319 | case SIOCOUTQNSD: |
6b96018b | 3320 | case SIOCATMARK: |
63ff03ab | 3321 | return sock_do_ioctl(net, sock, cmd, arg); |
9177efd3 AB |
3322 | } |
3323 | ||
6b96018b AB |
3324 | return -ENOIOCTLCMD; |
3325 | } | |
7a229387 | 3326 | |
95c96174 | 3327 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3328 | unsigned long arg) |
89bbfc95 SP |
3329 | { |
3330 | struct socket *sock = file->private_data; | |
3331 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3332 | struct sock *sk; |
3333 | struct net *net; | |
3334 | ||
3335 | sk = sock->sk; | |
3336 | net = sock_net(sk); | |
89bbfc95 SP |
3337 | |
3338 | if (sock->ops->compat_ioctl) | |
3339 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3340 | ||
87de87d5 DM |
3341 | if (ret == -ENOIOCTLCMD && |
3342 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3343 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3344 | ||
6b96018b AB |
3345 | if (ret == -ENOIOCTLCMD) |
3346 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3347 | ||
89bbfc95 SP |
3348 | return ret; |
3349 | } | |
3350 | #endif | |
3351 | ||
8a3c245c PT |
3352 | /** |
3353 | * kernel_bind - bind an address to a socket (kernel space) | |
3354 | * @sock: socket | |
3355 | * @addr: address | |
3356 | * @addrlen: length of address | |
3357 | * | |
3358 | * Returns 0 or an error. | |
3359 | */ | |
3360 | ||
ac5a488e SS |
3361 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3362 | { | |
3363 | return sock->ops->bind(sock, addr, addrlen); | |
3364 | } | |
c6d409cf | 3365 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e | 3366 | |
8a3c245c PT |
3367 | /** |
3368 | * kernel_listen - move socket to listening state (kernel space) | |
3369 | * @sock: socket | |
3370 | * @backlog: pending connections queue size | |
3371 | * | |
3372 | * Returns 0 or an error. | |
3373 | */ | |
3374 | ||
ac5a488e SS |
3375 | int kernel_listen(struct socket *sock, int backlog) |
3376 | { | |
3377 | return sock->ops->listen(sock, backlog); | |
3378 | } | |
c6d409cf | 3379 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e | 3380 | |
8a3c245c PT |
3381 | /** |
3382 | * kernel_accept - accept a connection (kernel space) | |
3383 | * @sock: listening socket | |
3384 | * @newsock: new connected socket | |
3385 | * @flags: flags | |
3386 | * | |
3387 | * @flags must be SOCK_CLOEXEC, SOCK_NONBLOCK or 0. | |
3388 | * If it fails, @newsock is guaranteed to be %NULL. | |
3389 | * Returns 0 or an error. | |
3390 | */ | |
3391 | ||
ac5a488e SS |
3392 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) |
3393 | { | |
3394 | struct sock *sk = sock->sk; | |
3395 | int err; | |
3396 | ||
3397 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3398 | newsock); | |
3399 | if (err < 0) | |
3400 | goto done; | |
3401 | ||
cdfbabfb | 3402 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3403 | if (err < 0) { |
3404 | sock_release(*newsock); | |
fa8705b0 | 3405 | *newsock = NULL; |
ac5a488e SS |
3406 | goto done; |
3407 | } | |
3408 | ||
3409 | (*newsock)->ops = sock->ops; | |
1b08534e | 3410 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3411 | |
3412 | done: | |
3413 | return err; | |
3414 | } | |
c6d409cf | 3415 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e | 3416 | |
8a3c245c PT |
3417 | /** |
3418 | * kernel_connect - connect a socket (kernel space) | |
3419 | * @sock: socket | |
3420 | * @addr: address | |
3421 | * @addrlen: address length | |
3422 | * @flags: flags (O_NONBLOCK, ...) | |
3423 | * | |
f1dcffcc | 3424 | * For datagram sockets, @addr is the address to which datagrams are sent |
8a3c245c PT |
3425 | * by default, and the only address from which datagrams are received. |
3426 | * For stream sockets, attempts to connect to @addr. | |
3427 | * Returns 0 or an error code. | |
3428 | */ | |
3429 | ||
ac5a488e | 3430 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, |
4768fbcb | 3431 | int flags) |
ac5a488e SS |
3432 | { |
3433 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3434 | } | |
c6d409cf | 3435 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e | 3436 | |
8a3c245c PT |
3437 | /** |
3438 | * kernel_getsockname - get the address which the socket is bound (kernel space) | |
3439 | * @sock: socket | |
3440 | * @addr: address holder | |
3441 | * | |
3442 | * Fills the @addr pointer with the address which the socket is bound. | |
3443 | * Returns 0 or an error code. | |
3444 | */ | |
3445 | ||
9b2c45d4 | 3446 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3447 | { |
9b2c45d4 | 3448 | return sock->ops->getname(sock, addr, 0); |
ac5a488e | 3449 | } |
c6d409cf | 3450 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e | 3451 | |
8a3c245c | 3452 | /** |
645f0897 | 3453 | * kernel_getpeername - get the address which the socket is connected (kernel space) |
8a3c245c PT |
3454 | * @sock: socket |
3455 | * @addr: address holder | |
3456 | * | |
3457 | * Fills the @addr pointer with the address which the socket is connected. | |
3458 | * Returns 0 or an error code. | |
3459 | */ | |
3460 | ||
9b2c45d4 | 3461 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3462 | { |
9b2c45d4 | 3463 | return sock->ops->getname(sock, addr, 1); |
ac5a488e | 3464 | } |
c6d409cf | 3465 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e | 3466 | |
8a3c245c PT |
3467 | /** |
3468 | * kernel_sendpage - send a &page through a socket (kernel space) | |
3469 | * @sock: socket | |
3470 | * @page: page | |
3471 | * @offset: page offset | |
3472 | * @size: total size in bytes | |
3473 | * @flags: flags (MSG_DONTWAIT, ...) | |
3474 | * | |
3475 | * Returns the total amount sent in bytes or an error. | |
3476 | */ | |
3477 | ||
ac5a488e SS |
3478 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, |
3479 | size_t size, int flags) | |
3480 | { | |
7b62d31d CL |
3481 | if (sock->ops->sendpage) { |
3482 | /* Warn in case the improper page to zero-copy send */ | |
3483 | WARN_ONCE(!sendpage_ok(page), "improper page for zero-copy send"); | |
ac5a488e | 3484 | return sock->ops->sendpage(sock, page, offset, size, flags); |
7b62d31d | 3485 | } |
ac5a488e SS |
3486 | return sock_no_sendpage(sock, page, offset, size, flags); |
3487 | } | |
c6d409cf | 3488 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3489 | |
8a3c245c PT |
3490 | /** |
3491 | * kernel_sendpage_locked - send a &page through the locked sock (kernel space) | |
3492 | * @sk: sock | |
3493 | * @page: page | |
3494 | * @offset: page offset | |
3495 | * @size: total size in bytes | |
3496 | * @flags: flags (MSG_DONTWAIT, ...) | |
3497 | * | |
3498 | * Returns the total amount sent in bytes or an error. | |
3499 | * Caller must hold @sk. | |
3500 | */ | |
3501 | ||
306b13eb TH |
3502 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3503 | size_t size, int flags) | |
3504 | { | |
3505 | struct socket *sock = sk->sk_socket; | |
3506 | ||
3507 | if (sock->ops->sendpage_locked) | |
3508 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3509 | flags); | |
3510 | ||
3511 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3512 | } | |
3513 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3514 | ||
8a3c245c | 3515 | /** |
645f0897 | 3516 | * kernel_sock_shutdown - shut down part of a full-duplex connection (kernel space) |
8a3c245c PT |
3517 | * @sock: socket |
3518 | * @how: connection part | |
3519 | * | |
3520 | * Returns 0 or an error. | |
3521 | */ | |
3522 | ||
91cf45f0 TM |
3523 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3524 | { | |
3525 | return sock->ops->shutdown(sock, how); | |
3526 | } | |
91cf45f0 | 3527 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 | 3528 | |
8a3c245c PT |
3529 | /** |
3530 | * kernel_sock_ip_overhead - returns the IP overhead imposed by a socket | |
3531 | * @sk: socket | |
3532 | * | |
3533 | * This routine returns the IP overhead imposed by a socket i.e. | |
3534 | * the length of the underlying IP header, depending on whether | |
3535 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
3536 | * on at the socket. Assumes that the caller has a lock on the socket. | |
113c3075 | 3537 | */ |
8a3c245c | 3538 | |
113c3075 P |
3539 | u32 kernel_sock_ip_overhead(struct sock *sk) |
3540 | { | |
3541 | struct inet_sock *inet; | |
3542 | struct ip_options_rcu *opt; | |
3543 | u32 overhead = 0; | |
113c3075 P |
3544 | #if IS_ENABLED(CONFIG_IPV6) |
3545 | struct ipv6_pinfo *np; | |
3546 | struct ipv6_txoptions *optv6 = NULL; | |
3547 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3548 | ||
3549 | if (!sk) | |
3550 | return overhead; | |
3551 | ||
113c3075 P |
3552 | switch (sk->sk_family) { |
3553 | case AF_INET: | |
3554 | inet = inet_sk(sk); | |
3555 | overhead += sizeof(struct iphdr); | |
3556 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3557 | sock_owned_by_user(sk)); |
113c3075 P |
3558 | if (opt) |
3559 | overhead += opt->opt.optlen; | |
3560 | return overhead; | |
3561 | #if IS_ENABLED(CONFIG_IPV6) | |
3562 | case AF_INET6: | |
3563 | np = inet6_sk(sk); | |
3564 | overhead += sizeof(struct ipv6hdr); | |
3565 | if (np) | |
3566 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3567 | sock_owned_by_user(sk)); |
113c3075 P |
3568 | if (optv6) |
3569 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3570 | return overhead; | |
3571 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3572 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3573 | return overhead; | |
3574 | } | |
3575 | } | |
3576 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |