]>
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 | ||
aef2feda | 55 | #include <linux/bpf-cgroup.h> |
cc69837f | 56 | #include <linux/ethtool.h> |
1da177e4 | 57 | #include <linux/mm.h> |
1da177e4 LT |
58 | #include <linux/socket.h> |
59 | #include <linux/file.h> | |
60 | #include <linux/net.h> | |
61 | #include <linux/interrupt.h> | |
aaca0bdc | 62 | #include <linux/thread_info.h> |
55737fda | 63 | #include <linux/rcupdate.h> |
1da177e4 LT |
64 | #include <linux/netdevice.h> |
65 | #include <linux/proc_fs.h> | |
66 | #include <linux/seq_file.h> | |
4a3e2f71 | 67 | #include <linux/mutex.h> |
1da177e4 | 68 | #include <linux/if_bridge.h> |
20380731 | 69 | #include <linux/if_vlan.h> |
408eccce | 70 | #include <linux/ptp_classify.h> |
1da177e4 LT |
71 | #include <linux/init.h> |
72 | #include <linux/poll.h> | |
73 | #include <linux/cache.h> | |
74 | #include <linux/module.h> | |
75 | #include <linux/highmem.h> | |
1da177e4 | 76 | #include <linux/mount.h> |
fba9be49 | 77 | #include <linux/pseudo_fs.h> |
1da177e4 LT |
78 | #include <linux/security.h> |
79 | #include <linux/syscalls.h> | |
80 | #include <linux/compat.h> | |
81 | #include <linux/kmod.h> | |
3ec3b2fb | 82 | #include <linux/audit.h> |
d86b5e0e | 83 | #include <linux/wireless.h> |
1b8d7ae4 | 84 | #include <linux/nsproxy.h> |
1fd7317d | 85 | #include <linux/magic.h> |
5a0e3ad6 | 86 | #include <linux/slab.h> |
600e1779 | 87 | #include <linux/xattr.h> |
c8e8cd57 | 88 | #include <linux/nospec.h> |
8c3c447b | 89 | #include <linux/indirect_call_wrapper.h> |
1da177e4 | 90 | |
7c0f6ba6 | 91 | #include <linux/uaccess.h> |
1da177e4 LT |
92 | #include <asm/unistd.h> |
93 | ||
94 | #include <net/compat.h> | |
87de87d5 | 95 | #include <net/wext.h> |
f8451725 | 96 | #include <net/cls_cgroup.h> |
1da177e4 LT |
97 | |
98 | #include <net/sock.h> | |
99 | #include <linux/netfilter.h> | |
100 | ||
6b96018b AB |
101 | #include <linux/if_tun.h> |
102 | #include <linux/ipv6_route.h> | |
103 | #include <linux/route.h> | |
c7dc504e | 104 | #include <linux/termios.h> |
6b96018b | 105 | #include <linux/sockios.h> |
076bb0c8 | 106 | #include <net/busy_poll.h> |
f24b9be5 | 107 | #include <linux/errqueue.h> |
d7c08826 | 108 | #include <linux/ptp_clock_kernel.h> |
06021292 | 109 | |
e0d1095a | 110 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
111 | unsigned int sysctl_net_busy_read __read_mostly; |
112 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 113 | #endif |
6b96018b | 114 | |
8ae5e030 AV |
115 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
116 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 117 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
118 | |
119 | static int sock_close(struct inode *inode, struct file *file); | |
a11e1d43 LT |
120 | static __poll_t sock_poll(struct file *file, |
121 | struct poll_table_struct *wait); | |
89bddce5 | 122 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
123 | #ifdef CONFIG_COMPAT |
124 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 125 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 126 | #endif |
1da177e4 | 127 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
128 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
129 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 130 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 131 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 132 | unsigned int flags); |
542d3065 AB |
133 | |
134 | #ifdef CONFIG_PROC_FS | |
135 | static void sock_show_fdinfo(struct seq_file *m, struct file *f) | |
136 | { | |
137 | struct socket *sock = f->private_data; | |
138 | ||
139 | if (sock->ops->show_fdinfo) | |
140 | sock->ops->show_fdinfo(m, sock); | |
141 | } | |
142 | #else | |
143 | #define sock_show_fdinfo NULL | |
144 | #endif | |
1da177e4 | 145 | |
1da177e4 LT |
146 | /* |
147 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
148 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
149 | */ | |
150 | ||
da7071d7 | 151 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
152 | .owner = THIS_MODULE, |
153 | .llseek = no_llseek, | |
8ae5e030 AV |
154 | .read_iter = sock_read_iter, |
155 | .write_iter = sock_write_iter, | |
1da177e4 LT |
156 | .poll = sock_poll, |
157 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
158 | #ifdef CONFIG_COMPAT |
159 | .compat_ioctl = compat_sock_ioctl, | |
160 | #endif | |
1da177e4 | 161 | .mmap = sock_mmap, |
1da177e4 LT |
162 | .release = sock_close, |
163 | .fasync = sock_fasync, | |
5274f052 JA |
164 | .sendpage = sock_sendpage, |
165 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 166 | .splice_read = sock_splice_read, |
b4653342 | 167 | .show_fdinfo = sock_show_fdinfo, |
1da177e4 LT |
168 | }; |
169 | ||
fe0bdbde YD |
170 | static const char * const pf_family_names[] = { |
171 | [PF_UNSPEC] = "PF_UNSPEC", | |
172 | [PF_UNIX] = "PF_UNIX/PF_LOCAL", | |
173 | [PF_INET] = "PF_INET", | |
174 | [PF_AX25] = "PF_AX25", | |
175 | [PF_IPX] = "PF_IPX", | |
176 | [PF_APPLETALK] = "PF_APPLETALK", | |
177 | [PF_NETROM] = "PF_NETROM", | |
178 | [PF_BRIDGE] = "PF_BRIDGE", | |
179 | [PF_ATMPVC] = "PF_ATMPVC", | |
180 | [PF_X25] = "PF_X25", | |
181 | [PF_INET6] = "PF_INET6", | |
182 | [PF_ROSE] = "PF_ROSE", | |
183 | [PF_DECnet] = "PF_DECnet", | |
184 | [PF_NETBEUI] = "PF_NETBEUI", | |
185 | [PF_SECURITY] = "PF_SECURITY", | |
186 | [PF_KEY] = "PF_KEY", | |
187 | [PF_NETLINK] = "PF_NETLINK/PF_ROUTE", | |
188 | [PF_PACKET] = "PF_PACKET", | |
189 | [PF_ASH] = "PF_ASH", | |
190 | [PF_ECONET] = "PF_ECONET", | |
191 | [PF_ATMSVC] = "PF_ATMSVC", | |
192 | [PF_RDS] = "PF_RDS", | |
193 | [PF_SNA] = "PF_SNA", | |
194 | [PF_IRDA] = "PF_IRDA", | |
195 | [PF_PPPOX] = "PF_PPPOX", | |
196 | [PF_WANPIPE] = "PF_WANPIPE", | |
197 | [PF_LLC] = "PF_LLC", | |
198 | [PF_IB] = "PF_IB", | |
199 | [PF_MPLS] = "PF_MPLS", | |
200 | [PF_CAN] = "PF_CAN", | |
201 | [PF_TIPC] = "PF_TIPC", | |
202 | [PF_BLUETOOTH] = "PF_BLUETOOTH", | |
203 | [PF_IUCV] = "PF_IUCV", | |
204 | [PF_RXRPC] = "PF_RXRPC", | |
205 | [PF_ISDN] = "PF_ISDN", | |
206 | [PF_PHONET] = "PF_PHONET", | |
207 | [PF_IEEE802154] = "PF_IEEE802154", | |
208 | [PF_CAIF] = "PF_CAIF", | |
209 | [PF_ALG] = "PF_ALG", | |
210 | [PF_NFC] = "PF_NFC", | |
211 | [PF_VSOCK] = "PF_VSOCK", | |
212 | [PF_KCM] = "PF_KCM", | |
213 | [PF_QIPCRTR] = "PF_QIPCRTR", | |
214 | [PF_SMC] = "PF_SMC", | |
215 | [PF_XDP] = "PF_XDP", | |
bc49d816 | 216 | [PF_MCTP] = "PF_MCTP", |
fe0bdbde YD |
217 | }; |
218 | ||
1da177e4 LT |
219 | /* |
220 | * The protocol list. Each protocol is registered in here. | |
221 | */ | |
222 | ||
1da177e4 | 223 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 224 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 225 | |
1da177e4 | 226 | /* |
89bddce5 SH |
227 | * Support routines. |
228 | * Move socket addresses back and forth across the kernel/user | |
229 | * divide and look after the messy bits. | |
1da177e4 LT |
230 | */ |
231 | ||
1da177e4 LT |
232 | /** |
233 | * move_addr_to_kernel - copy a socket address into kernel space | |
234 | * @uaddr: Address in user space | |
235 | * @kaddr: Address in kernel space | |
236 | * @ulen: Length in user space | |
237 | * | |
238 | * The address is copied into kernel space. If the provided address is | |
239 | * too long an error code of -EINVAL is returned. If the copy gives | |
240 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
241 | */ | |
242 | ||
43db362d | 243 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 244 | { |
230b1839 | 245 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 246 | return -EINVAL; |
89bddce5 | 247 | if (ulen == 0) |
1da177e4 | 248 | return 0; |
89bddce5 | 249 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 250 | return -EFAULT; |
3ec3b2fb | 251 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
252 | } |
253 | ||
254 | /** | |
255 | * move_addr_to_user - copy an address to user space | |
256 | * @kaddr: kernel space address | |
257 | * @klen: length of address in kernel | |
258 | * @uaddr: user space address | |
259 | * @ulen: pointer to user length field | |
260 | * | |
261 | * The value pointed to by ulen on entry is the buffer length available. | |
262 | * This is overwritten with the buffer space used. -EINVAL is returned | |
263 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
264 | * is returned if either the buffer or the length field are not | |
265 | * accessible. | |
266 | * After copying the data up to the limit the user specifies, the true | |
267 | * length of the data is written over the length limit the user | |
268 | * specified. Zero is returned for a success. | |
269 | */ | |
89bddce5 | 270 | |
43db362d | 271 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 272 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
273 | { |
274 | int err; | |
275 | int len; | |
276 | ||
68c6beb3 | 277 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
278 | err = get_user(len, ulen); |
279 | if (err) | |
1da177e4 | 280 | return err; |
89bddce5 SH |
281 | if (len > klen) |
282 | len = klen; | |
68c6beb3 | 283 | if (len < 0) |
1da177e4 | 284 | return -EINVAL; |
89bddce5 | 285 | if (len) { |
d6fe3945 SG |
286 | if (audit_sockaddr(klen, kaddr)) |
287 | return -ENOMEM; | |
89bddce5 | 288 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
289 | return -EFAULT; |
290 | } | |
291 | /* | |
89bddce5 SH |
292 | * "fromlen shall refer to the value before truncation.." |
293 | * 1003.1g | |
1da177e4 LT |
294 | */ |
295 | return __put_user(klen, ulen); | |
296 | } | |
297 | ||
08009a76 | 298 | static struct kmem_cache *sock_inode_cachep __ro_after_init; |
1da177e4 LT |
299 | |
300 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
301 | { | |
302 | struct socket_alloc *ei; | |
89bddce5 | 303 | |
fd60b288 | 304 | ei = alloc_inode_sb(sb, sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
305 | if (!ei) |
306 | return NULL; | |
333f7909 AV |
307 | init_waitqueue_head(&ei->socket.wq.wait); |
308 | ei->socket.wq.fasync_list = NULL; | |
309 | ei->socket.wq.flags = 0; | |
89bddce5 | 310 | |
1da177e4 LT |
311 | ei->socket.state = SS_UNCONNECTED; |
312 | ei->socket.flags = 0; | |
313 | ei->socket.ops = NULL; | |
314 | ei->socket.sk = NULL; | |
315 | ei->socket.file = NULL; | |
1da177e4 LT |
316 | |
317 | return &ei->vfs_inode; | |
318 | } | |
319 | ||
6d7855c5 | 320 | static void sock_free_inode(struct inode *inode) |
1da177e4 | 321 | { |
43815482 ED |
322 | struct socket_alloc *ei; |
323 | ||
324 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
43815482 | 325 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
326 | } |
327 | ||
51cc5068 | 328 | static void init_once(void *foo) |
1da177e4 | 329 | { |
89bddce5 | 330 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 331 | |
a35afb83 | 332 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 333 | } |
89bddce5 | 334 | |
1e911632 | 335 | static void init_inodecache(void) |
1da177e4 LT |
336 | { |
337 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
338 | sizeof(struct socket_alloc), |
339 | 0, | |
340 | (SLAB_HWCACHE_ALIGN | | |
341 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 342 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 343 | init_once); |
1e911632 | 344 | BUG_ON(sock_inode_cachep == NULL); |
1da177e4 LT |
345 | } |
346 | ||
b87221de | 347 | static const struct super_operations sockfs_ops = { |
c6d409cf | 348 | .alloc_inode = sock_alloc_inode, |
6d7855c5 | 349 | .free_inode = sock_free_inode, |
c6d409cf | 350 | .statfs = simple_statfs, |
1da177e4 LT |
351 | }; |
352 | ||
c23fbb6b ED |
353 | /* |
354 | * sockfs_dname() is called from d_path(). | |
355 | */ | |
356 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
357 | { | |
358 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 359 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
360 | } |
361 | ||
3ba13d17 | 362 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 363 | .d_dname = sockfs_dname, |
1da177e4 LT |
364 | }; |
365 | ||
bba0bd31 AG |
366 | static int sockfs_xattr_get(const struct xattr_handler *handler, |
367 | struct dentry *dentry, struct inode *inode, | |
368 | const char *suffix, void *value, size_t size) | |
369 | { | |
370 | if (value) { | |
371 | if (dentry->d_name.len + 1 > size) | |
372 | return -ERANGE; | |
373 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
374 | } | |
375 | return dentry->d_name.len + 1; | |
376 | } | |
377 | ||
378 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" | |
379 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
380 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
381 | ||
382 | static const struct xattr_handler sockfs_xattr_handler = { | |
383 | .name = XATTR_NAME_SOCKPROTONAME, | |
384 | .get = sockfs_xattr_get, | |
385 | }; | |
386 | ||
4a590153 | 387 | static int sockfs_security_xattr_set(const struct xattr_handler *handler, |
e65ce2a5 | 388 | struct user_namespace *mnt_userns, |
4a590153 AG |
389 | struct dentry *dentry, struct inode *inode, |
390 | const char *suffix, const void *value, | |
391 | size_t size, int flags) | |
392 | { | |
393 | /* Handled by LSM. */ | |
394 | return -EAGAIN; | |
395 | } | |
396 | ||
397 | static const struct xattr_handler sockfs_security_xattr_handler = { | |
398 | .prefix = XATTR_SECURITY_PREFIX, | |
399 | .set = sockfs_security_xattr_set, | |
400 | }; | |
401 | ||
bba0bd31 AG |
402 | static const struct xattr_handler *sockfs_xattr_handlers[] = { |
403 | &sockfs_xattr_handler, | |
4a590153 | 404 | &sockfs_security_xattr_handler, |
bba0bd31 AG |
405 | NULL |
406 | }; | |
407 | ||
fba9be49 | 408 | static int sockfs_init_fs_context(struct fs_context *fc) |
c74a1cbb | 409 | { |
fba9be49 DH |
410 | struct pseudo_fs_context *ctx = init_pseudo(fc, SOCKFS_MAGIC); |
411 | if (!ctx) | |
412 | return -ENOMEM; | |
413 | ctx->ops = &sockfs_ops; | |
414 | ctx->dops = &sockfs_dentry_operations; | |
415 | ctx->xattr = sockfs_xattr_handlers; | |
416 | return 0; | |
c74a1cbb AV |
417 | } |
418 | ||
419 | static struct vfsmount *sock_mnt __read_mostly; | |
420 | ||
421 | static struct file_system_type sock_fs_type = { | |
422 | .name = "sockfs", | |
fba9be49 | 423 | .init_fs_context = sockfs_init_fs_context, |
c74a1cbb AV |
424 | .kill_sb = kill_anon_super, |
425 | }; | |
426 | ||
1da177e4 LT |
427 | /* |
428 | * Obtains the first available file descriptor and sets it up for use. | |
429 | * | |
39d8c1b6 DM |
430 | * These functions create file structures and maps them to fd space |
431 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
432 | * and file struct implicitly stored in sock->file. |
433 | * Note that another thread may close file descriptor before we return | |
434 | * from this function. We use the fact that now we do not refer | |
435 | * to socket after mapping. If one day we will need it, this | |
436 | * function will increment ref. count on file by 1. | |
437 | * | |
438 | * In any case returned fd MAY BE not valid! | |
439 | * This race condition is unavoidable | |
440 | * with shared fd spaces, we cannot solve it inside kernel, | |
441 | * but we take care of internal coherence yet. | |
442 | */ | |
443 | ||
8a3c245c PT |
444 | /** |
445 | * sock_alloc_file - Bind a &socket to a &file | |
446 | * @sock: socket | |
447 | * @flags: file status flags | |
448 | * @dname: protocol name | |
449 | * | |
450 | * Returns the &file bound with @sock, implicitly storing it | |
451 | * in sock->file. If dname is %NULL, sets to "". | |
452 | * On failure the return is a ERR pointer (see linux/err.h). | |
453 | * This function uses GFP_KERNEL internally. | |
454 | */ | |
455 | ||
aab174f0 | 456 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 457 | { |
7cbe66b6 | 458 | struct file *file; |
1da177e4 | 459 | |
d93aa9d8 AV |
460 | if (!dname) |
461 | dname = sock->sk ? sock->sk->sk_prot_creator->name : ""; | |
39d8c1b6 | 462 | |
d93aa9d8 AV |
463 | file = alloc_file_pseudo(SOCK_INODE(sock), sock_mnt, dname, |
464 | O_RDWR | (flags & O_NONBLOCK), | |
465 | &socket_file_ops); | |
b5ffe634 | 466 | if (IS_ERR(file)) { |
8e1611e2 | 467 | sock_release(sock); |
39b65252 | 468 | return file; |
cc3808f8 AV |
469 | } |
470 | ||
471 | sock->file = file; | |
39d8c1b6 | 472 | file->private_data = sock; |
d8e464ec | 473 | stream_open(SOCK_INODE(sock), file); |
28407630 | 474 | return file; |
39d8c1b6 | 475 | } |
56b31d1c | 476 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 477 | |
56b31d1c | 478 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
479 | { |
480 | struct file *newfile; | |
28407630 | 481 | int fd = get_unused_fd_flags(flags); |
ce4bb04c AV |
482 | if (unlikely(fd < 0)) { |
483 | sock_release(sock); | |
28407630 | 484 | return fd; |
ce4bb04c | 485 | } |
39d8c1b6 | 486 | |
aab174f0 | 487 | newfile = sock_alloc_file(sock, flags, NULL); |
4546e44c | 488 | if (!IS_ERR(newfile)) { |
39d8c1b6 | 489 | fd_install(fd, newfile); |
28407630 AV |
490 | return fd; |
491 | } | |
7cbe66b6 | 492 | |
28407630 AV |
493 | put_unused_fd(fd); |
494 | return PTR_ERR(newfile); | |
1da177e4 LT |
495 | } |
496 | ||
8a3c245c PT |
497 | /** |
498 | * sock_from_file - Return the &socket bounded to @file. | |
499 | * @file: file | |
8a3c245c | 500 | * |
dba4a925 | 501 | * On failure returns %NULL. |
8a3c245c PT |
502 | */ |
503 | ||
dba4a925 | 504 | struct socket *sock_from_file(struct file *file) |
6cb153ca | 505 | { |
6cb153ca | 506 | if (file->f_op == &socket_file_ops) |
da214a47 | 507 | return file->private_data; /* set in sock_alloc_file */ |
6cb153ca | 508 | |
23bb80d2 | 509 | return NULL; |
6cb153ca | 510 | } |
406a3c63 | 511 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 512 | |
1da177e4 | 513 | /** |
c6d409cf | 514 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
515 | * @fd: file handle |
516 | * @err: pointer to an error code return | |
517 | * | |
518 | * The file handle passed in is locked and the socket it is bound | |
241c4667 | 519 | * to is returned. If an error occurs the err pointer is overwritten |
1da177e4 LT |
520 | * with a negative errno code and NULL is returned. The function checks |
521 | * for both invalid handles and passing a handle which is not a socket. | |
522 | * | |
523 | * On a success the socket object pointer is returned. | |
524 | */ | |
525 | ||
526 | struct socket *sockfd_lookup(int fd, int *err) | |
527 | { | |
528 | struct file *file; | |
1da177e4 LT |
529 | struct socket *sock; |
530 | ||
89bddce5 SH |
531 | file = fget(fd); |
532 | if (!file) { | |
1da177e4 LT |
533 | *err = -EBADF; |
534 | return NULL; | |
535 | } | |
89bddce5 | 536 | |
dba4a925 FR |
537 | sock = sock_from_file(file); |
538 | if (!sock) { | |
539 | *err = -ENOTSOCK; | |
1da177e4 | 540 | fput(file); |
dba4a925 | 541 | } |
6cb153ca BL |
542 | return sock; |
543 | } | |
c6d409cf | 544 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 545 | |
6cb153ca BL |
546 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
547 | { | |
00e188ef | 548 | struct fd f = fdget(fd); |
6cb153ca BL |
549 | struct socket *sock; |
550 | ||
3672558c | 551 | *err = -EBADF; |
00e188ef | 552 | if (f.file) { |
dba4a925 | 553 | sock = sock_from_file(f.file); |
00e188ef | 554 | if (likely(sock)) { |
ce787a5a | 555 | *fput_needed = f.flags & FDPUT_FPUT; |
6cb153ca | 556 | return sock; |
00e188ef | 557 | } |
dba4a925 | 558 | *err = -ENOTSOCK; |
00e188ef | 559 | fdput(f); |
1da177e4 | 560 | } |
6cb153ca | 561 | return NULL; |
1da177e4 LT |
562 | } |
563 | ||
600e1779 MY |
564 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, |
565 | size_t size) | |
566 | { | |
567 | ssize_t len; | |
568 | ssize_t used = 0; | |
569 | ||
c5ef6035 | 570 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
571 | if (len < 0) |
572 | return len; | |
573 | used += len; | |
574 | if (buffer) { | |
575 | if (size < used) | |
576 | return -ERANGE; | |
577 | buffer += len; | |
578 | } | |
579 | ||
580 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
581 | used += len; | |
582 | if (buffer) { | |
583 | if (size < used) | |
584 | return -ERANGE; | |
585 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
586 | buffer += len; | |
587 | } | |
588 | ||
589 | return used; | |
590 | } | |
591 | ||
549c7297 CB |
592 | static int sockfs_setattr(struct user_namespace *mnt_userns, |
593 | struct dentry *dentry, struct iattr *iattr) | |
86741ec2 | 594 | { |
549c7297 | 595 | int err = simple_setattr(&init_user_ns, dentry, iattr); |
86741ec2 | 596 | |
e1a3a60a | 597 | if (!err && (iattr->ia_valid & ATTR_UID)) { |
86741ec2 LC |
598 | struct socket *sock = SOCKET_I(d_inode(dentry)); |
599 | ||
6d8c50dc CW |
600 | if (sock->sk) |
601 | sock->sk->sk_uid = iattr->ia_uid; | |
602 | else | |
603 | err = -ENOENT; | |
86741ec2 LC |
604 | } |
605 | ||
606 | return err; | |
607 | } | |
608 | ||
600e1779 | 609 | static const struct inode_operations sockfs_inode_ops = { |
600e1779 | 610 | .listxattr = sockfs_listxattr, |
86741ec2 | 611 | .setattr = sockfs_setattr, |
600e1779 MY |
612 | }; |
613 | ||
1da177e4 | 614 | /** |
8a3c245c | 615 | * sock_alloc - allocate a socket |
89bddce5 | 616 | * |
1da177e4 LT |
617 | * Allocate a new inode and socket object. The two are bound together |
618 | * and initialised. The socket is then returned. If we are out of inodes | |
8a3c245c | 619 | * NULL is returned. This functions uses GFP_KERNEL internally. |
1da177e4 LT |
620 | */ |
621 | ||
f4a00aac | 622 | struct socket *sock_alloc(void) |
1da177e4 | 623 | { |
89bddce5 SH |
624 | struct inode *inode; |
625 | struct socket *sock; | |
1da177e4 | 626 | |
a209dfc7 | 627 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
628 | if (!inode) |
629 | return NULL; | |
630 | ||
631 | sock = SOCKET_I(inode); | |
632 | ||
85fe4025 | 633 | inode->i_ino = get_next_ino(); |
89bddce5 | 634 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
635 | inode->i_uid = current_fsuid(); |
636 | inode->i_gid = current_fsgid(); | |
600e1779 | 637 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 638 | |
1da177e4 LT |
639 | return sock; |
640 | } | |
f4a00aac | 641 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 642 | |
6d8c50dc | 643 | static void __sock_release(struct socket *sock, struct inode *inode) |
1da177e4 LT |
644 | { |
645 | if (sock->ops) { | |
646 | struct module *owner = sock->ops->owner; | |
647 | ||
6d8c50dc CW |
648 | if (inode) |
649 | inode_lock(inode); | |
1da177e4 | 650 | sock->ops->release(sock); |
ff7b11aa | 651 | sock->sk = NULL; |
6d8c50dc CW |
652 | if (inode) |
653 | inode_unlock(inode); | |
1da177e4 LT |
654 | sock->ops = NULL; |
655 | module_put(owner); | |
656 | } | |
657 | ||
333f7909 | 658 | if (sock->wq.fasync_list) |
3410f22e | 659 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 660 | |
1da177e4 LT |
661 | if (!sock->file) { |
662 | iput(SOCK_INODE(sock)); | |
663 | return; | |
664 | } | |
89bddce5 | 665 | sock->file = NULL; |
1da177e4 | 666 | } |
6d8c50dc | 667 | |
9a8ad9ac AL |
668 | /** |
669 | * sock_release - close a socket | |
670 | * @sock: socket to close | |
671 | * | |
672 | * The socket is released from the protocol stack if it has a release | |
673 | * callback, and the inode is then released if the socket is bound to | |
674 | * an inode not a file. | |
675 | */ | |
6d8c50dc CW |
676 | void sock_release(struct socket *sock) |
677 | { | |
678 | __sock_release(sock, NULL); | |
679 | } | |
c6d409cf | 680 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 681 | |
c14ac945 | 682 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 683 | { |
140c55d4 ED |
684 | u8 flags = *tx_flags; |
685 | ||
51eb7492 | 686 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) { |
140c55d4 ED |
687 | flags |= SKBTX_HW_TSTAMP; |
688 | ||
51eb7492 GE |
689 | /* PTP hardware clocks can provide a free running cycle counter |
690 | * as a time base for virtual clocks. Tell driver to use the | |
691 | * free running cycle counter for timestamp if socket is bound | |
692 | * to virtual clock. | |
693 | */ | |
694 | if (tsflags & SOF_TIMESTAMPING_BIND_PHC) | |
695 | flags |= SKBTX_HW_TSTAMP_USE_CYCLES; | |
696 | } | |
697 | ||
c14ac945 | 698 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
699 | flags |= SKBTX_SW_TSTAMP; |
700 | ||
c14ac945 | 701 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
702 | flags |= SKBTX_SCHED_TSTAMP; |
703 | ||
140c55d4 | 704 | *tx_flags = flags; |
20d49473 | 705 | } |
67cc0d40 | 706 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 707 | |
8c3c447b PA |
708 | INDIRECT_CALLABLE_DECLARE(int inet_sendmsg(struct socket *, struct msghdr *, |
709 | size_t)); | |
a648a592 PA |
710 | INDIRECT_CALLABLE_DECLARE(int inet6_sendmsg(struct socket *, struct msghdr *, |
711 | size_t)); | |
d8725c86 | 712 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 713 | { |
a648a592 PA |
714 | int ret = INDIRECT_CALL_INET(sock->ops->sendmsg, inet6_sendmsg, |
715 | inet_sendmsg, sock, msg, | |
716 | msg_data_left(msg)); | |
d8725c86 AV |
717 | BUG_ON(ret == -EIOCBQUEUED); |
718 | return ret; | |
1da177e4 LT |
719 | } |
720 | ||
85806af0 RD |
721 | /** |
722 | * sock_sendmsg - send a message through @sock | |
723 | * @sock: socket | |
724 | * @msg: message to send | |
725 | * | |
726 | * Sends @msg through @sock, passing through LSM. | |
727 | * Returns the number of bytes sent, or an error code. | |
728 | */ | |
d8725c86 | 729 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 730 | { |
d8725c86 | 731 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 732 | msg_data_left(msg)); |
228e548e | 733 | |
d8725c86 | 734 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 735 | } |
c6d409cf | 736 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 | 737 | |
8a3c245c PT |
738 | /** |
739 | * kernel_sendmsg - send a message through @sock (kernel-space) | |
740 | * @sock: socket | |
741 | * @msg: message header | |
742 | * @vec: kernel vec | |
743 | * @num: vec array length | |
744 | * @size: total message data size | |
745 | * | |
746 | * Builds the message data with @vec and sends it through @sock. | |
747 | * Returns the number of bytes sent, or an error code. | |
748 | */ | |
749 | ||
1da177e4 LT |
750 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
751 | struct kvec *vec, size_t num, size_t size) | |
752 | { | |
aa563d7b | 753 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
d8725c86 | 754 | return sock_sendmsg(sock, msg); |
1da177e4 | 755 | } |
c6d409cf | 756 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 757 | |
8a3c245c PT |
758 | /** |
759 | * kernel_sendmsg_locked - send a message through @sock (kernel-space) | |
760 | * @sk: sock | |
761 | * @msg: message header | |
762 | * @vec: output s/g array | |
763 | * @num: output s/g array length | |
764 | * @size: total message data size | |
765 | * | |
766 | * Builds the message data with @vec and sends it through @sock. | |
767 | * Returns the number of bytes sent, or an error code. | |
768 | * Caller must hold @sk. | |
769 | */ | |
770 | ||
306b13eb TH |
771 | int kernel_sendmsg_locked(struct sock *sk, struct msghdr *msg, |
772 | struct kvec *vec, size_t num, size_t size) | |
773 | { | |
774 | struct socket *sock = sk->sk_socket; | |
775 | ||
776 | if (!sock->ops->sendmsg_locked) | |
db5980d8 | 777 | return sock_no_sendmsg_locked(sk, msg, size); |
306b13eb | 778 | |
aa563d7b | 779 | iov_iter_kvec(&msg->msg_iter, WRITE, vec, num, size); |
306b13eb TH |
780 | |
781 | return sock->ops->sendmsg_locked(sk, msg, msg_data_left(msg)); | |
782 | } | |
783 | EXPORT_SYMBOL(kernel_sendmsg_locked); | |
784 | ||
8605330a SHY |
785 | static bool skb_is_err_queue(const struct sk_buff *skb) |
786 | { | |
787 | /* pkt_type of skbs enqueued on the error queue are set to | |
788 | * PACKET_OUTGOING in skb_set_err_queue(). This is only safe to do | |
789 | * in recvmsg, since skbs received on a local socket will never | |
790 | * have a pkt_type of PACKET_OUTGOING. | |
791 | */ | |
792 | return skb->pkt_type == PACKET_OUTGOING; | |
793 | } | |
794 | ||
b50a5c70 ML |
795 | /* On transmit, software and hardware timestamps are returned independently. |
796 | * As the two skb clones share the hardware timestamp, which may be updated | |
797 | * before the software timestamp is received, a hardware TX timestamp may be | |
798 | * returned only if there is no software TX timestamp. Ignore false software | |
799 | * timestamps, which may be made in the __sock_recv_timestamp() call when the | |
7f1bc6e9 | 800 | * option SO_TIMESTAMP_OLD(NS) is enabled on the socket, even when the skb has a |
b50a5c70 ML |
801 | * hardware timestamp. |
802 | */ | |
803 | static bool skb_is_swtx_tstamp(const struct sk_buff *skb, int false_tstamp) | |
804 | { | |
805 | return skb->tstamp && !false_tstamp && skb_is_err_queue(skb); | |
806 | } | |
807 | ||
97dc7cd9 GE |
808 | static ktime_t get_timestamp(struct sock *sk, struct sk_buff *skb, int *if_index) |
809 | { | |
810 | bool cycles = sk->sk_tsflags & SOF_TIMESTAMPING_BIND_PHC; | |
811 | struct skb_shared_hwtstamps *shhwtstamps = skb_hwtstamps(skb); | |
812 | struct net_device *orig_dev; | |
813 | ktime_t hwtstamp; | |
814 | ||
815 | rcu_read_lock(); | |
816 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
817 | if (orig_dev) { | |
818 | *if_index = orig_dev->ifindex; | |
819 | hwtstamp = netdev_get_tstamp(orig_dev, shhwtstamps, cycles); | |
820 | } else { | |
821 | hwtstamp = shhwtstamps->hwtstamp; | |
822 | } | |
823 | rcu_read_unlock(); | |
824 | ||
825 | return hwtstamp; | |
826 | } | |
827 | ||
828 | static void put_ts_pktinfo(struct msghdr *msg, struct sk_buff *skb, | |
829 | int if_index) | |
aad9c8c4 ML |
830 | { |
831 | struct scm_ts_pktinfo ts_pktinfo; | |
832 | struct net_device *orig_dev; | |
833 | ||
834 | if (!skb_mac_header_was_set(skb)) | |
835 | return; | |
836 | ||
837 | memset(&ts_pktinfo, 0, sizeof(ts_pktinfo)); | |
838 | ||
97dc7cd9 GE |
839 | if (!if_index) { |
840 | rcu_read_lock(); | |
841 | orig_dev = dev_get_by_napi_id(skb_napi_id(skb)); | |
842 | if (orig_dev) | |
843 | if_index = orig_dev->ifindex; | |
844 | rcu_read_unlock(); | |
845 | } | |
846 | ts_pktinfo.if_index = if_index; | |
aad9c8c4 ML |
847 | |
848 | ts_pktinfo.pkt_length = skb->len - skb_mac_offset(skb); | |
849 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_PKTINFO, | |
850 | sizeof(ts_pktinfo), &ts_pktinfo); | |
851 | } | |
852 | ||
92f37fd2 ED |
853 | /* |
854 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
855 | */ | |
856 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
857 | struct sk_buff *skb) | |
858 | { | |
20d49473 | 859 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
887feae3 | 860 | int new_tstamp = sock_flag(sk, SOCK_TSTAMP_NEW); |
9718475e DD |
861 | struct scm_timestamping_internal tss; |
862 | ||
b50a5c70 | 863 | int empty = 1, false_tstamp = 0; |
20d49473 PO |
864 | struct skb_shared_hwtstamps *shhwtstamps = |
865 | skb_hwtstamps(skb); | |
97dc7cd9 | 866 | int if_index; |
007747a9 | 867 | ktime_t hwtstamp; |
20d49473 PO |
868 | |
869 | /* Race occurred between timestamp enabling and packet | |
870 | receiving. Fill in the current time for now. */ | |
b50a5c70 | 871 | if (need_software_tstamp && skb->tstamp == 0) { |
20d49473 | 872 | __net_timestamp(skb); |
b50a5c70 ML |
873 | false_tstamp = 1; |
874 | } | |
20d49473 PO |
875 | |
876 | if (need_software_tstamp) { | |
877 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
887feae3 DD |
878 | if (new_tstamp) { |
879 | struct __kernel_sock_timeval tv; | |
880 | ||
881 | skb_get_new_timestamp(skb, &tv); | |
882 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_NEW, | |
883 | sizeof(tv), &tv); | |
884 | } else { | |
885 | struct __kernel_old_timeval tv; | |
886 | ||
887 | skb_get_timestamp(skb, &tv); | |
888 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMP_OLD, | |
889 | sizeof(tv), &tv); | |
890 | } | |
20d49473 | 891 | } else { |
887feae3 DD |
892 | if (new_tstamp) { |
893 | struct __kernel_timespec ts; | |
894 | ||
895 | skb_get_new_timestampns(skb, &ts); | |
896 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_NEW, | |
897 | sizeof(ts), &ts); | |
898 | } else { | |
df1b4ba9 | 899 | struct __kernel_old_timespec ts; |
887feae3 DD |
900 | |
901 | skb_get_timestampns(skb, &ts); | |
902 | put_cmsg(msg, SOL_SOCKET, SO_TIMESTAMPNS_OLD, | |
903 | sizeof(ts), &ts); | |
904 | } | |
20d49473 PO |
905 | } |
906 | } | |
907 | ||
f24b9be5 | 908 | memset(&tss, 0, sizeof(tss)); |
c199105d | 909 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
9718475e | 910 | ktime_to_timespec64_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 911 | empty = 0; |
4d276eb6 | 912 | if (shhwtstamps && |
b9f40e21 | 913 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
d7c08826 | 914 | !skb_is_swtx_tstamp(skb, false_tstamp)) { |
97dc7cd9 GE |
915 | if_index = 0; |
916 | if (skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP_NETDEV) | |
917 | hwtstamp = get_timestamp(sk, skb, &if_index); | |
007747a9 ML |
918 | else |
919 | hwtstamp = shhwtstamps->hwtstamp; | |
d7c08826 | 920 | |
97dc7cd9 GE |
921 | if (sk->sk_tsflags & SOF_TIMESTAMPING_BIND_PHC) |
922 | hwtstamp = ptp_convert_timestamp(&hwtstamp, | |
923 | sk->sk_bind_phc); | |
924 | ||
007747a9 | 925 | if (ktime_to_timespec64_cond(hwtstamp, tss.ts + 2)) { |
d7c08826 YL |
926 | empty = 0; |
927 | ||
928 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_OPT_PKTINFO) && | |
929 | !skb_is_err_queue(skb)) | |
97dc7cd9 | 930 | put_ts_pktinfo(msg, skb, if_index); |
d7c08826 | 931 | } |
aad9c8c4 | 932 | } |
1c885808 | 933 | if (!empty) { |
9718475e DD |
934 | if (sock_flag(sk, SOCK_TSTAMP_NEW)) |
935 | put_cmsg_scm_timestamping64(msg, &tss); | |
936 | else | |
937 | put_cmsg_scm_timestamping(msg, &tss); | |
1c885808 | 938 | |
8605330a | 939 | if (skb_is_err_queue(skb) && skb->len && |
4ef1b286 | 940 | SKB_EXT_ERR(skb)->opt_stats) |
1c885808 FY |
941 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPING_OPT_STATS, |
942 | skb->len, skb->data); | |
943 | } | |
92f37fd2 | 944 | } |
7c81fd8b ACM |
945 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
946 | ||
6e3e939f JB |
947 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
948 | struct sk_buff *skb) | |
949 | { | |
950 | int ack; | |
951 | ||
952 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
953 | return; | |
954 | if (!skb->wifi_acked_valid) | |
955 | return; | |
956 | ||
957 | ack = skb->wifi_acked; | |
958 | ||
959 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
960 | } | |
961 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
962 | ||
11165f14 | 963 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
964 | struct sk_buff *skb) | |
3b885787 | 965 | { |
744d5a3e | 966 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 967 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 968 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
969 | } |
970 | ||
6fd1d51c EM |
971 | static void sock_recv_mark(struct msghdr *msg, struct sock *sk, |
972 | struct sk_buff *skb) | |
973 | { | |
974 | if (sock_flag(sk, SOCK_RCVMARK) && skb) | |
975 | put_cmsg(msg, SOL_SOCKET, SO_MARK, sizeof(__u32), | |
976 | &skb->mark); | |
977 | } | |
978 | ||
979 | void __sock_recv_cmsgs(struct msghdr *msg, struct sock *sk, | |
980 | struct sk_buff *skb) | |
3b885787 NH |
981 | { |
982 | sock_recv_timestamp(msg, sk, skb); | |
983 | sock_recv_drops(msg, sk, skb); | |
6fd1d51c | 984 | sock_recv_mark(msg, sk, skb); |
3b885787 | 985 | } |
6fd1d51c | 986 | EXPORT_SYMBOL_GPL(__sock_recv_cmsgs); |
3b885787 | 987 | |
8c3c447b | 988 | INDIRECT_CALLABLE_DECLARE(int inet_recvmsg(struct socket *, struct msghdr *, |
a648a592 PA |
989 | size_t, int)); |
990 | INDIRECT_CALLABLE_DECLARE(int inet6_recvmsg(struct socket *, struct msghdr *, | |
991 | size_t, int)); | |
1b784140 | 992 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 993 | int flags) |
1da177e4 | 994 | { |
a648a592 PA |
995 | return INDIRECT_CALL_INET(sock->ops->recvmsg, inet6_recvmsg, |
996 | inet_recvmsg, sock, msg, msg_data_left(msg), | |
997 | flags); | |
1da177e4 LT |
998 | } |
999 | ||
85806af0 RD |
1000 | /** |
1001 | * sock_recvmsg - receive a message from @sock | |
1002 | * @sock: socket | |
1003 | * @msg: message to receive | |
1004 | * @flags: message flags | |
1005 | * | |
1006 | * Receives @msg from @sock, passing through LSM. Returns the total number | |
1007 | * of bytes received, or an error. | |
1008 | */ | |
2da62906 | 1009 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 1010 | { |
2da62906 | 1011 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 1012 | |
2da62906 | 1013 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 1014 | } |
c6d409cf | 1015 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 1016 | |
c1249c0a | 1017 | /** |
8a3c245c PT |
1018 | * kernel_recvmsg - Receive a message from a socket (kernel space) |
1019 | * @sock: The socket to receive the message from | |
1020 | * @msg: Received message | |
1021 | * @vec: Input s/g array for message data | |
1022 | * @num: Size of input s/g array | |
1023 | * @size: Number of bytes to read | |
1024 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
c1249c0a | 1025 | * |
8a3c245c PT |
1026 | * On return the msg structure contains the scatter/gather array passed in the |
1027 | * vec argument. The array is modified so that it consists of the unfilled | |
1028 | * portion of the original array. | |
c1249c0a | 1029 | * |
8a3c245c | 1030 | * The returned value is the total number of bytes received, or an error. |
c1249c0a | 1031 | */ |
8a3c245c | 1032 | |
89bddce5 SH |
1033 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
1034 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 | 1035 | { |
1f466e1f | 1036 | msg->msg_control_is_user = false; |
aa563d7b | 1037 | iov_iter_kvec(&msg->msg_iter, READ, vec, num, size); |
1f466e1f | 1038 | return sock_recvmsg(sock, msg, flags); |
1da177e4 | 1039 | } |
c6d409cf | 1040 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 1041 | |
ce1d4d3e CH |
1042 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
1043 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 1044 | { |
1da177e4 LT |
1045 | struct socket *sock; |
1046 | int flags; | |
1047 | ||
ce1d4d3e CH |
1048 | sock = file->private_data; |
1049 | ||
35f9c09f ED |
1050 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
1051 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
1052 | flags |= more; | |
ce1d4d3e | 1053 | |
e6949583 | 1054 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 1055 | } |
1da177e4 | 1056 | |
9c55e01c | 1057 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 1058 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
1059 | unsigned int flags) |
1060 | { | |
1061 | struct socket *sock = file->private_data; | |
1062 | ||
997b37da | 1063 | if (unlikely(!sock->ops->splice_read)) |
95506588 | 1064 | return generic_file_splice_read(file, ppos, pipe, len, flags); |
997b37da | 1065 | |
9c55e01c JA |
1066 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
1067 | } | |
1068 | ||
8ae5e030 | 1069 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 1070 | { |
6d652330 AV |
1071 | struct file *file = iocb->ki_filp; |
1072 | struct socket *sock = file->private_data; | |
0345f931 | 1073 | struct msghdr msg = {.msg_iter = *to, |
1074 | .msg_iocb = iocb}; | |
8ae5e030 | 1075 | ssize_t res; |
ce1d4d3e | 1076 | |
ebfcd895 | 1077 | if (file->f_flags & O_NONBLOCK || (iocb->ki_flags & IOCB_NOWAIT)) |
8ae5e030 AV |
1078 | msg.msg_flags = MSG_DONTWAIT; |
1079 | ||
1080 | if (iocb->ki_pos != 0) | |
1da177e4 | 1081 | return -ESPIPE; |
027445c3 | 1082 | |
66ee59af | 1083 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
1084 | return 0; |
1085 | ||
2da62906 | 1086 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
1087 | *to = msg.msg_iter; |
1088 | return res; | |
1da177e4 LT |
1089 | } |
1090 | ||
8ae5e030 | 1091 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 1092 | { |
6d652330 AV |
1093 | struct file *file = iocb->ki_filp; |
1094 | struct socket *sock = file->private_data; | |
0345f931 | 1095 | struct msghdr msg = {.msg_iter = *from, |
1096 | .msg_iocb = iocb}; | |
8ae5e030 | 1097 | ssize_t res; |
1da177e4 | 1098 | |
8ae5e030 | 1099 | if (iocb->ki_pos != 0) |
ce1d4d3e | 1100 | return -ESPIPE; |
027445c3 | 1101 | |
ebfcd895 | 1102 | if (file->f_flags & O_NONBLOCK || (iocb->ki_flags & IOCB_NOWAIT)) |
8ae5e030 AV |
1103 | msg.msg_flags = MSG_DONTWAIT; |
1104 | ||
6d652330 AV |
1105 | if (sock->type == SOCK_SEQPACKET) |
1106 | msg.msg_flags |= MSG_EOR; | |
1107 | ||
d8725c86 | 1108 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
1109 | *from = msg.msg_iter; |
1110 | return res; | |
1da177e4 LT |
1111 | } |
1112 | ||
1da177e4 LT |
1113 | /* |
1114 | * Atomic setting of ioctl hooks to avoid race | |
1115 | * with module unload. | |
1116 | */ | |
1117 | ||
4a3e2f71 | 1118 | static DEFINE_MUTEX(br_ioctl_mutex); |
ad2f99ae AB |
1119 | static int (*br_ioctl_hook)(struct net *net, struct net_bridge *br, |
1120 | unsigned int cmd, struct ifreq *ifr, | |
1121 | void __user *uarg); | |
1da177e4 | 1122 | |
ad2f99ae AB |
1123 | void brioctl_set(int (*hook)(struct net *net, struct net_bridge *br, |
1124 | unsigned int cmd, struct ifreq *ifr, | |
1125 | void __user *uarg)) | |
1da177e4 | 1126 | { |
4a3e2f71 | 1127 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 1128 | br_ioctl_hook = hook; |
4a3e2f71 | 1129 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1130 | } |
1131 | EXPORT_SYMBOL(brioctl_set); | |
1132 | ||
ad2f99ae AB |
1133 | int br_ioctl_call(struct net *net, struct net_bridge *br, unsigned int cmd, |
1134 | struct ifreq *ifr, void __user *uarg) | |
1135 | { | |
1136 | int err = -ENOPKG; | |
1137 | ||
1138 | if (!br_ioctl_hook) | |
1139 | request_module("bridge"); | |
1140 | ||
1141 | mutex_lock(&br_ioctl_mutex); | |
1142 | if (br_ioctl_hook) | |
1143 | err = br_ioctl_hook(net, br, cmd, ifr, uarg); | |
1144 | mutex_unlock(&br_ioctl_mutex); | |
1145 | ||
1146 | return err; | |
1147 | } | |
1148 | ||
4a3e2f71 | 1149 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 1150 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 1151 | |
881d966b | 1152 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 1153 | { |
4a3e2f71 | 1154 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1155 | vlan_ioctl_hook = hook; |
4a3e2f71 | 1156 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
1157 | } |
1158 | EXPORT_SYMBOL(vlan_ioctl_set); | |
1159 | ||
6b96018b | 1160 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
63ff03ab | 1161 | unsigned int cmd, unsigned long arg) |
6b96018b | 1162 | { |
876f0bf9 AB |
1163 | struct ifreq ifr; |
1164 | bool need_copyout; | |
6b96018b AB |
1165 | int err; |
1166 | void __user *argp = (void __user *)arg; | |
a554bf96 | 1167 | void __user *data; |
6b96018b AB |
1168 | |
1169 | err = sock->ops->ioctl(sock, cmd, arg); | |
1170 | ||
1171 | /* | |
1172 | * If this ioctl is unknown try to hand it down | |
1173 | * to the NIC driver. | |
1174 | */ | |
36fd633e AV |
1175 | if (err != -ENOIOCTLCMD) |
1176 | return err; | |
6b96018b | 1177 | |
29ce8f97 JK |
1178 | if (!is_socket_ioctl_cmd(cmd)) |
1179 | return -ENOTTY; | |
1180 | ||
a554bf96 | 1181 | if (get_user_ifreq(&ifr, &data, argp)) |
876f0bf9 | 1182 | return -EFAULT; |
a554bf96 | 1183 | err = dev_ioctl(net, cmd, &ifr, data, &need_copyout); |
876f0bf9 | 1184 | if (!err && need_copyout) |
a554bf96 | 1185 | if (put_user_ifreq(&ifr, argp)) |
44c02a2c | 1186 | return -EFAULT; |
876f0bf9 | 1187 | |
6b96018b AB |
1188 | return err; |
1189 | } | |
1190 | ||
1da177e4 LT |
1191 | /* |
1192 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1193 | * what to do with it - that's up to the protocol still. | |
1194 | */ | |
1195 | ||
1196 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) | |
1197 | { | |
1198 | struct socket *sock; | |
881d966b | 1199 | struct sock *sk; |
1da177e4 LT |
1200 | void __user *argp = (void __user *)arg; |
1201 | int pid, err; | |
881d966b | 1202 | struct net *net; |
1da177e4 | 1203 | |
b69aee04 | 1204 | sock = file->private_data; |
881d966b | 1205 | sk = sock->sk; |
3b1e0a65 | 1206 | net = sock_net(sk); |
44c02a2c AV |
1207 | if (unlikely(cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))) { |
1208 | struct ifreq ifr; | |
a554bf96 | 1209 | void __user *data; |
44c02a2c | 1210 | bool need_copyout; |
a554bf96 | 1211 | if (get_user_ifreq(&ifr, &data, argp)) |
44c02a2c | 1212 | return -EFAULT; |
a554bf96 | 1213 | err = dev_ioctl(net, cmd, &ifr, data, &need_copyout); |
44c02a2c | 1214 | if (!err && need_copyout) |
a554bf96 | 1215 | if (put_user_ifreq(&ifr, argp)) |
44c02a2c | 1216 | return -EFAULT; |
1da177e4 | 1217 | } else |
3d23e349 | 1218 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1219 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
b1b0c245 | 1220 | err = wext_handle_ioctl(net, cmd, argp); |
1da177e4 | 1221 | } else |
3d23e349 | 1222 | #endif |
89bddce5 | 1223 | switch (cmd) { |
1da177e4 LT |
1224 | case FIOSETOWN: |
1225 | case SIOCSPGRP: | |
1226 | err = -EFAULT; | |
1227 | if (get_user(pid, (int __user *)argp)) | |
1228 | break; | |
393cc3f5 | 1229 | err = f_setown(sock->file, pid, 1); |
1da177e4 LT |
1230 | break; |
1231 | case FIOGETOWN: | |
1232 | case SIOCGPGRP: | |
609d7fa9 | 1233 | err = put_user(f_getown(sock->file), |
89bddce5 | 1234 | (int __user *)argp); |
1da177e4 LT |
1235 | break; |
1236 | case SIOCGIFBR: | |
1237 | case SIOCSIFBR: | |
1238 | case SIOCBRADDBR: | |
1239 | case SIOCBRDELBR: | |
ad2f99ae | 1240 | err = br_ioctl_call(net, NULL, cmd, NULL, argp); |
1da177e4 LT |
1241 | break; |
1242 | case SIOCGIFVLAN: | |
1243 | case SIOCSIFVLAN: | |
1244 | err = -ENOPKG; | |
1245 | if (!vlan_ioctl_hook) | |
1246 | request_module("8021q"); | |
1247 | ||
4a3e2f71 | 1248 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1249 | if (vlan_ioctl_hook) |
881d966b | 1250 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1251 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1252 | break; |
c62cce2c AV |
1253 | case SIOCGSKNS: |
1254 | err = -EPERM; | |
1255 | if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) | |
1256 | break; | |
1257 | ||
1258 | err = open_related_ns(&net->ns, get_net_ns); | |
1259 | break; | |
0768e170 AB |
1260 | case SIOCGSTAMP_OLD: |
1261 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
1262 | if (!sock->ops->gettstamp) { |
1263 | err = -ENOIOCTLCMD; | |
1264 | break; | |
1265 | } | |
1266 | err = sock->ops->gettstamp(sock, argp, | |
0768e170 AB |
1267 | cmd == SIOCGSTAMP_OLD, |
1268 | !IS_ENABLED(CONFIG_64BIT)); | |
60747828 | 1269 | break; |
0768e170 AB |
1270 | case SIOCGSTAMP_NEW: |
1271 | case SIOCGSTAMPNS_NEW: | |
1272 | if (!sock->ops->gettstamp) { | |
1273 | err = -ENOIOCTLCMD; | |
1274 | break; | |
1275 | } | |
1276 | err = sock->ops->gettstamp(sock, argp, | |
1277 | cmd == SIOCGSTAMP_NEW, | |
1278 | false); | |
c7cbdbf2 | 1279 | break; |
876f0bf9 AB |
1280 | |
1281 | case SIOCGIFCONF: | |
1282 | err = dev_ifconf(net, argp); | |
1283 | break; | |
1284 | ||
1da177e4 | 1285 | default: |
63ff03ab | 1286 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1287 | break; |
89bddce5 | 1288 | } |
1da177e4 LT |
1289 | return err; |
1290 | } | |
1291 | ||
8a3c245c PT |
1292 | /** |
1293 | * sock_create_lite - creates a socket | |
1294 | * @family: protocol family (AF_INET, ...) | |
1295 | * @type: communication type (SOCK_STREAM, ...) | |
1296 | * @protocol: protocol (0, ...) | |
1297 | * @res: new socket | |
1298 | * | |
1299 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1300 | * The new socket initialization is not complete, see kernel_accept(). | |
1301 | * Returns 0 or an error. On failure @res is set to %NULL. | |
1302 | * This function internally uses GFP_KERNEL. | |
1303 | */ | |
1304 | ||
1da177e4 LT |
1305 | int sock_create_lite(int family, int type, int protocol, struct socket **res) |
1306 | { | |
1307 | int err; | |
1308 | struct socket *sock = NULL; | |
89bddce5 | 1309 | |
1da177e4 LT |
1310 | err = security_socket_create(family, type, protocol, 1); |
1311 | if (err) | |
1312 | goto out; | |
1313 | ||
1314 | sock = sock_alloc(); | |
1315 | if (!sock) { | |
1316 | err = -ENOMEM; | |
1317 | goto out; | |
1318 | } | |
1319 | ||
1da177e4 | 1320 | sock->type = type; |
7420ed23 VY |
1321 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1322 | if (err) | |
1323 | goto out_release; | |
1324 | ||
1da177e4 LT |
1325 | out: |
1326 | *res = sock; | |
1327 | return err; | |
7420ed23 VY |
1328 | out_release: |
1329 | sock_release(sock); | |
1330 | sock = NULL; | |
1331 | goto out; | |
1da177e4 | 1332 | } |
c6d409cf | 1333 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1334 | |
1335 | /* No kernel lock held - perfect */ | |
ade994f4 | 1336 | static __poll_t sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1337 | { |
3cafb376 | 1338 | struct socket *sock = file->private_data; |
a331de3b | 1339 | __poll_t events = poll_requested_events(wait), flag = 0; |
2d48d67f | 1340 | |
e88958e6 CH |
1341 | if (!sock->ops->poll) |
1342 | return 0; | |
f641f13b | 1343 | |
a331de3b CH |
1344 | if (sk_can_busy_loop(sock->sk)) { |
1345 | /* poll once if requested by the syscall */ | |
1346 | if (events & POLL_BUSY_LOOP) | |
1347 | sk_busy_loop(sock->sk, 1); | |
1348 | ||
1349 | /* if this socket can poll_ll, tell the system call */ | |
1350 | flag = POLL_BUSY_LOOP; | |
1351 | } | |
1352 | ||
1353 | return sock->ops->poll(file, sock, wait) | flag; | |
1da177e4 LT |
1354 | } |
1355 | ||
89bddce5 | 1356 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1357 | { |
b69aee04 | 1358 | struct socket *sock = file->private_data; |
1da177e4 LT |
1359 | |
1360 | return sock->ops->mmap(file, sock, vma); | |
1361 | } | |
1362 | ||
20380731 | 1363 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1364 | { |
6d8c50dc | 1365 | __sock_release(SOCKET_I(inode), inode); |
1da177e4 LT |
1366 | return 0; |
1367 | } | |
1368 | ||
1369 | /* | |
1370 | * Update the socket async list | |
1371 | * | |
1372 | * Fasync_list locking strategy. | |
1373 | * | |
1374 | * 1. fasync_list is modified only under process context socket lock | |
1375 | * i.e. under semaphore. | |
1376 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1377 | * or under socket lock |
1da177e4 LT |
1378 | */ |
1379 | ||
1380 | static int sock_fasync(int fd, struct file *filp, int on) | |
1381 | { | |
989a2979 ED |
1382 | struct socket *sock = filp->private_data; |
1383 | struct sock *sk = sock->sk; | |
333f7909 | 1384 | struct socket_wq *wq = &sock->wq; |
1da177e4 | 1385 | |
989a2979 | 1386 | if (sk == NULL) |
1da177e4 | 1387 | return -EINVAL; |
1da177e4 LT |
1388 | |
1389 | lock_sock(sk); | |
eaefd110 | 1390 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1391 | |
eaefd110 | 1392 | if (!wq->fasync_list) |
989a2979 ED |
1393 | sock_reset_flag(sk, SOCK_FASYNC); |
1394 | else | |
bcdce719 | 1395 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1396 | |
989a2979 | 1397 | release_sock(sk); |
1da177e4 LT |
1398 | return 0; |
1399 | } | |
1400 | ||
ceb5d58b | 1401 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1402 | |
ceb5d58b | 1403 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1404 | { |
ceb5d58b | 1405 | if (!wq || !wq->fasync_list) |
1da177e4 | 1406 | return -1; |
ceb5d58b | 1407 | |
89bddce5 | 1408 | switch (how) { |
8d8ad9d7 | 1409 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1410 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1411 | break; |
1412 | goto call_kill; | |
8d8ad9d7 | 1413 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1414 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 | 1415 | break; |
7c7ab580 | 1416 | fallthrough; |
8d8ad9d7 | 1417 | case SOCK_WAKE_IO: |
89bddce5 | 1418 | call_kill: |
43815482 | 1419 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1420 | break; |
8d8ad9d7 | 1421 | case SOCK_WAKE_URG: |
43815482 | 1422 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1423 | } |
ceb5d58b | 1424 | |
1da177e4 LT |
1425 | return 0; |
1426 | } | |
c6d409cf | 1427 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1428 | |
8a3c245c PT |
1429 | /** |
1430 | * __sock_create - creates a socket | |
1431 | * @net: net namespace | |
1432 | * @family: protocol family (AF_INET, ...) | |
1433 | * @type: communication type (SOCK_STREAM, ...) | |
1434 | * @protocol: protocol (0, ...) | |
1435 | * @res: new socket | |
1436 | * @kern: boolean for kernel space sockets | |
1437 | * | |
1438 | * Creates a new socket and assigns it to @res, passing through LSM. | |
1439 | * Returns 0 or an error. On failure @res is set to %NULL. @kern must | |
1440 | * be set to true if the socket resides in kernel space. | |
1441 | * This function internally uses GFP_KERNEL. | |
1442 | */ | |
1443 | ||
721db93a | 1444 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1445 | struct socket **res, int kern) |
1da177e4 LT |
1446 | { |
1447 | int err; | |
1448 | struct socket *sock; | |
55737fda | 1449 | const struct net_proto_family *pf; |
1da177e4 LT |
1450 | |
1451 | /* | |
89bddce5 | 1452 | * Check protocol is in range |
1da177e4 LT |
1453 | */ |
1454 | if (family < 0 || family >= NPROTO) | |
1455 | return -EAFNOSUPPORT; | |
1456 | if (type < 0 || type >= SOCK_MAX) | |
1457 | return -EINVAL; | |
1458 | ||
1459 | /* Compatibility. | |
1460 | ||
1461 | This uglymoron is moved from INET layer to here to avoid | |
1462 | deadlock in module load. | |
1463 | */ | |
1464 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1465 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1466 | current->comm); | |
1da177e4 LT |
1467 | family = PF_PACKET; |
1468 | } | |
1469 | ||
1470 | err = security_socket_create(family, type, protocol, kern); | |
1471 | if (err) | |
1472 | return err; | |
89bddce5 | 1473 | |
55737fda SH |
1474 | /* |
1475 | * Allocate the socket and allow the family to set things up. if | |
1476 | * the protocol is 0, the family is instructed to select an appropriate | |
1477 | * default. | |
1478 | */ | |
1479 | sock = sock_alloc(); | |
1480 | if (!sock) { | |
e87cc472 | 1481 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1482 | return -ENFILE; /* Not exactly a match, but its the |
1483 | closest posix thing */ | |
1484 | } | |
1485 | ||
1486 | sock->type = type; | |
1487 | ||
95a5afca | 1488 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1489 | /* Attempt to load a protocol module if the find failed. |
1490 | * | |
1491 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1492 | * requested real, full-featured networking support upon configuration. |
1493 | * Otherwise module support will break! | |
1494 | */ | |
190683a9 | 1495 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1496 | request_module("net-pf-%d", family); |
1da177e4 LT |
1497 | #endif |
1498 | ||
55737fda SH |
1499 | rcu_read_lock(); |
1500 | pf = rcu_dereference(net_families[family]); | |
1501 | err = -EAFNOSUPPORT; | |
1502 | if (!pf) | |
1503 | goto out_release; | |
1da177e4 LT |
1504 | |
1505 | /* | |
1506 | * We will call the ->create function, that possibly is in a loadable | |
1507 | * module, so we have to bump that loadable module refcnt first. | |
1508 | */ | |
55737fda | 1509 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1510 | goto out_release; |
1511 | ||
55737fda SH |
1512 | /* Now protected by module ref count */ |
1513 | rcu_read_unlock(); | |
1514 | ||
3f378b68 | 1515 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1516 | if (err < 0) |
1da177e4 | 1517 | goto out_module_put; |
a79af59e | 1518 | |
1da177e4 LT |
1519 | /* |
1520 | * Now to bump the refcnt of the [loadable] module that owns this | |
1521 | * socket at sock_release time we decrement its refcnt. | |
1522 | */ | |
55737fda SH |
1523 | if (!try_module_get(sock->ops->owner)) |
1524 | goto out_module_busy; | |
1525 | ||
1da177e4 LT |
1526 | /* |
1527 | * Now that we're done with the ->create function, the [loadable] | |
1528 | * module can have its refcnt decremented | |
1529 | */ | |
55737fda | 1530 | module_put(pf->owner); |
7420ed23 VY |
1531 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1532 | if (err) | |
3b185525 | 1533 | goto out_sock_release; |
55737fda | 1534 | *res = sock; |
1da177e4 | 1535 | |
55737fda SH |
1536 | return 0; |
1537 | ||
1538 | out_module_busy: | |
1539 | err = -EAFNOSUPPORT; | |
1da177e4 | 1540 | out_module_put: |
55737fda SH |
1541 | sock->ops = NULL; |
1542 | module_put(pf->owner); | |
1543 | out_sock_release: | |
1da177e4 | 1544 | sock_release(sock); |
55737fda SH |
1545 | return err; |
1546 | ||
1547 | out_release: | |
1548 | rcu_read_unlock(); | |
1549 | goto out_sock_release; | |
1da177e4 | 1550 | } |
721db93a | 1551 | EXPORT_SYMBOL(__sock_create); |
1da177e4 | 1552 | |
8a3c245c PT |
1553 | /** |
1554 | * sock_create - creates a socket | |
1555 | * @family: protocol family (AF_INET, ...) | |
1556 | * @type: communication type (SOCK_STREAM, ...) | |
1557 | * @protocol: protocol (0, ...) | |
1558 | * @res: new socket | |
1559 | * | |
1560 | * A wrapper around __sock_create(). | |
1561 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1562 | */ | |
1563 | ||
1da177e4 LT |
1564 | int sock_create(int family, int type, int protocol, struct socket **res) |
1565 | { | |
1b8d7ae4 | 1566 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1567 | } |
c6d409cf | 1568 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1569 | |
8a3c245c PT |
1570 | /** |
1571 | * sock_create_kern - creates a socket (kernel space) | |
1572 | * @net: net namespace | |
1573 | * @family: protocol family (AF_INET, ...) | |
1574 | * @type: communication type (SOCK_STREAM, ...) | |
1575 | * @protocol: protocol (0, ...) | |
1576 | * @res: new socket | |
1577 | * | |
1578 | * A wrapper around __sock_create(). | |
1579 | * Returns 0 or an error. This function internally uses GFP_KERNEL. | |
1580 | */ | |
1581 | ||
eeb1bd5c | 1582 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1583 | { |
eeb1bd5c | 1584 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1585 | } |
c6d409cf | 1586 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1587 | |
da214a47 | 1588 | static struct socket *__sys_socket_create(int family, int type, int protocol) |
1da177e4 | 1589 | { |
1da177e4 | 1590 | struct socket *sock; |
da214a47 | 1591 | int retval; |
a677a039 | 1592 | |
e38b36f3 UD |
1593 | /* Check the SOCK_* constants for consistency. */ |
1594 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1595 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1596 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1597 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1598 | ||
da214a47 JA |
1599 | if ((type & ~SOCK_TYPE_MASK) & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
1600 | return ERR_PTR(-EINVAL); | |
a677a039 | 1601 | type &= SOCK_TYPE_MASK; |
1da177e4 | 1602 | |
da214a47 JA |
1603 | retval = sock_create(family, type, protocol, &sock); |
1604 | if (retval < 0) | |
1605 | return ERR_PTR(retval); | |
1606 | ||
1607 | return sock; | |
1608 | } | |
1609 | ||
1610 | struct file *__sys_socket_file(int family, int type, int protocol) | |
1611 | { | |
1612 | struct socket *sock; | |
1613 | struct file *file; | |
1614 | int flags; | |
1615 | ||
1616 | sock = __sys_socket_create(family, type, protocol); | |
1617 | if (IS_ERR(sock)) | |
1618 | return ERR_CAST(sock); | |
1619 | ||
1620 | flags = type & ~SOCK_TYPE_MASK; | |
aaca0bdc UD |
1621 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1622 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1623 | ||
da214a47 JA |
1624 | file = sock_alloc_file(sock, flags, NULL); |
1625 | if (IS_ERR(file)) | |
1626 | sock_release(sock); | |
1627 | ||
1628 | return file; | |
1629 | } | |
1630 | ||
1631 | int __sys_socket(int family, int type, int protocol) | |
1632 | { | |
1633 | struct socket *sock; | |
1634 | int flags; | |
1635 | ||
1636 | sock = __sys_socket_create(family, type, protocol); | |
1637 | if (IS_ERR(sock)) | |
1638 | return PTR_ERR(sock); | |
1639 | ||
1640 | flags = type & ~SOCK_TYPE_MASK; | |
1641 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1642 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1da177e4 | 1643 | |
8e1611e2 | 1644 | return sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1645 | } |
1646 | ||
9d6a15c3 DB |
1647 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1648 | { | |
1649 | return __sys_socket(family, type, protocol); | |
1650 | } | |
1651 | ||
1da177e4 LT |
1652 | /* |
1653 | * Create a pair of connected sockets. | |
1654 | */ | |
1655 | ||
6debc8d8 | 1656 | int __sys_socketpair(int family, int type, int protocol, int __user *usockvec) |
1da177e4 LT |
1657 | { |
1658 | struct socket *sock1, *sock2; | |
1659 | int fd1, fd2, err; | |
db349509 | 1660 | struct file *newfile1, *newfile2; |
a677a039 UD |
1661 | int flags; |
1662 | ||
1663 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1664 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1665 | return -EINVAL; |
1666 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1667 | |
aaca0bdc UD |
1668 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1669 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1670 | ||
016a266b AV |
1671 | /* |
1672 | * reserve descriptors and make sure we won't fail | |
1673 | * to return them to userland. | |
1674 | */ | |
1675 | fd1 = get_unused_fd_flags(flags); | |
1676 | if (unlikely(fd1 < 0)) | |
1677 | return fd1; | |
1678 | ||
1679 | fd2 = get_unused_fd_flags(flags); | |
1680 | if (unlikely(fd2 < 0)) { | |
1681 | put_unused_fd(fd1); | |
1682 | return fd2; | |
1683 | } | |
1684 | ||
1685 | err = put_user(fd1, &usockvec[0]); | |
1686 | if (err) | |
1687 | goto out; | |
1688 | ||
1689 | err = put_user(fd2, &usockvec[1]); | |
1690 | if (err) | |
1691 | goto out; | |
1692 | ||
1da177e4 LT |
1693 | /* |
1694 | * Obtain the first socket and check if the underlying protocol | |
1695 | * supports the socketpair call. | |
1696 | */ | |
1697 | ||
1698 | err = sock_create(family, type, protocol, &sock1); | |
016a266b | 1699 | if (unlikely(err < 0)) |
1da177e4 LT |
1700 | goto out; |
1701 | ||
1702 | err = sock_create(family, type, protocol, &sock2); | |
016a266b AV |
1703 | if (unlikely(err < 0)) { |
1704 | sock_release(sock1); | |
1705 | goto out; | |
bf3c23d1 | 1706 | } |
d73aa286 | 1707 | |
d47cd945 DH |
1708 | err = security_socket_socketpair(sock1, sock2); |
1709 | if (unlikely(err)) { | |
1710 | sock_release(sock2); | |
1711 | sock_release(sock1); | |
1712 | goto out; | |
1713 | } | |
1714 | ||
016a266b AV |
1715 | err = sock1->ops->socketpair(sock1, sock2); |
1716 | if (unlikely(err < 0)) { | |
1717 | sock_release(sock2); | |
1718 | sock_release(sock1); | |
1719 | goto out; | |
28407630 AV |
1720 | } |
1721 | ||
aab174f0 | 1722 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1723 | if (IS_ERR(newfile1)) { |
28407630 | 1724 | err = PTR_ERR(newfile1); |
016a266b AV |
1725 | sock_release(sock2); |
1726 | goto out; | |
28407630 AV |
1727 | } |
1728 | ||
aab174f0 | 1729 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1730 | if (IS_ERR(newfile2)) { |
1731 | err = PTR_ERR(newfile2); | |
016a266b AV |
1732 | fput(newfile1); |
1733 | goto out; | |
db349509 AV |
1734 | } |
1735 | ||
157cf649 | 1736 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1737 | |
db349509 AV |
1738 | fd_install(fd1, newfile1); |
1739 | fd_install(fd2, newfile2); | |
d73aa286 | 1740 | return 0; |
1da177e4 | 1741 | |
016a266b | 1742 | out: |
d73aa286 | 1743 | put_unused_fd(fd2); |
d73aa286 | 1744 | put_unused_fd(fd1); |
1da177e4 LT |
1745 | return err; |
1746 | } | |
1747 | ||
6debc8d8 DB |
1748 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1749 | int __user *, usockvec) | |
1750 | { | |
1751 | return __sys_socketpair(family, type, protocol, usockvec); | |
1752 | } | |
1753 | ||
1da177e4 LT |
1754 | /* |
1755 | * Bind a name to a socket. Nothing much to do here since it's | |
1756 | * the protocol's responsibility to handle the local address. | |
1757 | * | |
1758 | * We move the socket address to kernel space before we call | |
1759 | * the protocol layer (having also checked the address is ok). | |
1760 | */ | |
1761 | ||
a87d35d8 | 1762 | int __sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
1da177e4 LT |
1763 | { |
1764 | struct socket *sock; | |
230b1839 | 1765 | struct sockaddr_storage address; |
6cb153ca | 1766 | int err, fput_needed; |
1da177e4 | 1767 | |
89bddce5 | 1768 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1769 | if (sock) { |
43db362d | 1770 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
068b88cc | 1771 | if (!err) { |
89bddce5 | 1772 | err = security_socket_bind(sock, |
230b1839 | 1773 | (struct sockaddr *)&address, |
89bddce5 | 1774 | addrlen); |
6cb153ca BL |
1775 | if (!err) |
1776 | err = sock->ops->bind(sock, | |
89bddce5 | 1777 | (struct sockaddr *) |
230b1839 | 1778 | &address, addrlen); |
1da177e4 | 1779 | } |
6cb153ca | 1780 | fput_light(sock->file, fput_needed); |
89bddce5 | 1781 | } |
1da177e4 LT |
1782 | return err; |
1783 | } | |
1784 | ||
a87d35d8 DB |
1785 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1786 | { | |
1787 | return __sys_bind(fd, umyaddr, addrlen); | |
1788 | } | |
1789 | ||
1da177e4 LT |
1790 | /* |
1791 | * Perform a listen. Basically, we allow the protocol to do anything | |
1792 | * necessary for a listen, and if that works, we mark the socket as | |
1793 | * ready for listening. | |
1794 | */ | |
1795 | ||
25e290ee | 1796 | int __sys_listen(int fd, int backlog) |
1da177e4 LT |
1797 | { |
1798 | struct socket *sock; | |
6cb153ca | 1799 | int err, fput_needed; |
b8e1f9b5 | 1800 | int somaxconn; |
89bddce5 SH |
1801 | |
1802 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1803 | if (sock) { | |
8efa6e93 | 1804 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1805 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1806 | backlog = somaxconn; |
1da177e4 LT |
1807 | |
1808 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1809 | if (!err) |
1810 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1811 | |
6cb153ca | 1812 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1813 | } |
1814 | return err; | |
1815 | } | |
1816 | ||
25e290ee DB |
1817 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1818 | { | |
1819 | return __sys_listen(fd, backlog); | |
1820 | } | |
1821 | ||
d32f89da | 1822 | struct file *do_accept(struct file *file, unsigned file_flags, |
de2ea4b6 | 1823 | struct sockaddr __user *upeer_sockaddr, |
d32f89da | 1824 | int __user *upeer_addrlen, int flags) |
1da177e4 LT |
1825 | { |
1826 | struct socket *sock, *newsock; | |
39d8c1b6 | 1827 | struct file *newfile; |
d32f89da | 1828 | int err, len; |
230b1839 | 1829 | struct sockaddr_storage address; |
1da177e4 | 1830 | |
dba4a925 | 1831 | sock = sock_from_file(file); |
d32f89da PB |
1832 | if (!sock) |
1833 | return ERR_PTR(-ENOTSOCK); | |
1da177e4 | 1834 | |
c6d409cf ED |
1835 | newsock = sock_alloc(); |
1836 | if (!newsock) | |
d32f89da | 1837 | return ERR_PTR(-ENFILE); |
1da177e4 LT |
1838 | |
1839 | newsock->type = sock->type; | |
1840 | newsock->ops = sock->ops; | |
1841 | ||
1da177e4 LT |
1842 | /* |
1843 | * We don't need try_module_get here, as the listening socket (sock) | |
1844 | * has the protocol module (sock->ops->owner) held. | |
1845 | */ | |
1846 | __module_get(newsock->ops->owner); | |
1847 | ||
aab174f0 | 1848 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
d32f89da PB |
1849 | if (IS_ERR(newfile)) |
1850 | return newfile; | |
39d8c1b6 | 1851 | |
a79af59e FF |
1852 | err = security_socket_accept(sock, newsock); |
1853 | if (err) | |
39d8c1b6 | 1854 | goto out_fd; |
a79af59e | 1855 | |
de2ea4b6 JA |
1856 | err = sock->ops->accept(sock, newsock, sock->file->f_flags | file_flags, |
1857 | false); | |
1da177e4 | 1858 | if (err < 0) |
39d8c1b6 | 1859 | goto out_fd; |
1da177e4 LT |
1860 | |
1861 | if (upeer_sockaddr) { | |
9b2c45d4 DV |
1862 | len = newsock->ops->getname(newsock, |
1863 | (struct sockaddr *)&address, 2); | |
1864 | if (len < 0) { | |
1da177e4 | 1865 | err = -ECONNABORTED; |
39d8c1b6 | 1866 | goto out_fd; |
1da177e4 | 1867 | } |
43db362d | 1868 | err = move_addr_to_user(&address, |
230b1839 | 1869 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1870 | if (err < 0) |
39d8c1b6 | 1871 | goto out_fd; |
1da177e4 LT |
1872 | } |
1873 | ||
1874 | /* File flags are not inherited via accept() unlike another OSes. */ | |
d32f89da | 1875 | return newfile; |
39d8c1b6 | 1876 | out_fd: |
9606a216 | 1877 | fput(newfile); |
d32f89da PB |
1878 | return ERR_PTR(err); |
1879 | } | |
1880 | ||
1881 | int __sys_accept4_file(struct file *file, unsigned file_flags, | |
1882 | struct sockaddr __user *upeer_sockaddr, | |
1883 | int __user *upeer_addrlen, int flags, | |
1884 | unsigned long nofile) | |
1885 | { | |
1886 | struct file *newfile; | |
1887 | int newfd; | |
1888 | ||
1889 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) | |
1890 | return -EINVAL; | |
1891 | ||
1892 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1893 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
de2ea4b6 | 1894 | |
d32f89da PB |
1895 | newfd = __get_unused_fd_flags(flags, nofile); |
1896 | if (unlikely(newfd < 0)) | |
1897 | return newfd; | |
1898 | ||
1899 | newfile = do_accept(file, file_flags, upeer_sockaddr, upeer_addrlen, | |
1900 | flags); | |
1901 | if (IS_ERR(newfile)) { | |
1902 | put_unused_fd(newfd); | |
1903 | return PTR_ERR(newfile); | |
1904 | } | |
1905 | fd_install(newfd, newfile); | |
1906 | return newfd; | |
de2ea4b6 JA |
1907 | } |
1908 | ||
1909 | /* | |
1910 | * For accept, we attempt to create a new socket, set up the link | |
1911 | * with the client, wake up the client, then return the new | |
1912 | * connected fd. We collect the address of the connector in kernel | |
1913 | * space and move it to user at the very end. This is unclean because | |
1914 | * we open the socket then return an error. | |
1915 | * | |
1916 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1917 | * status to recvmsg. We need to add that support in a way thats | |
1918 | * clean when we restructure accept also. | |
1919 | */ | |
1920 | ||
1921 | int __sys_accept4(int fd, struct sockaddr __user *upeer_sockaddr, | |
1922 | int __user *upeer_addrlen, int flags) | |
1923 | { | |
1924 | int ret = -EBADF; | |
1925 | struct fd f; | |
1926 | ||
1927 | f = fdget(fd); | |
1928 | if (f.file) { | |
1929 | ret = __sys_accept4_file(f.file, 0, upeer_sockaddr, | |
09952e3e JA |
1930 | upeer_addrlen, flags, |
1931 | rlimit(RLIMIT_NOFILE)); | |
6b07edeb | 1932 | fdput(f); |
de2ea4b6 JA |
1933 | } |
1934 | ||
1935 | return ret; | |
1da177e4 LT |
1936 | } |
1937 | ||
4541e805 DB |
1938 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1939 | int __user *, upeer_addrlen, int, flags) | |
1940 | { | |
1941 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, flags); | |
1942 | } | |
1943 | ||
20f37034 HC |
1944 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1945 | int __user *, upeer_addrlen) | |
aaca0bdc | 1946 | { |
4541e805 | 1947 | return __sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1948 | } |
1949 | ||
1da177e4 LT |
1950 | /* |
1951 | * Attempt to connect to a socket with the server address. The address | |
1952 | * is in user space so we verify it is OK and move it to kernel space. | |
1953 | * | |
1954 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1955 | * break bindings | |
1956 | * | |
1957 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1958 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1959 | * include the -EINPROGRESS status for such sockets. | |
1960 | */ | |
1961 | ||
f499a021 | 1962 | int __sys_connect_file(struct file *file, struct sockaddr_storage *address, |
bd3ded31 | 1963 | int addrlen, int file_flags) |
1da177e4 LT |
1964 | { |
1965 | struct socket *sock; | |
bd3ded31 | 1966 | int err; |
1da177e4 | 1967 | |
dba4a925 FR |
1968 | sock = sock_from_file(file); |
1969 | if (!sock) { | |
1970 | err = -ENOTSOCK; | |
1da177e4 | 1971 | goto out; |
dba4a925 | 1972 | } |
1da177e4 | 1973 | |
89bddce5 | 1974 | err = |
f499a021 | 1975 | security_socket_connect(sock, (struct sockaddr *)address, addrlen); |
1da177e4 | 1976 | if (err) |
bd3ded31 | 1977 | goto out; |
1da177e4 | 1978 | |
f499a021 | 1979 | err = sock->ops->connect(sock, (struct sockaddr *)address, addrlen, |
bd3ded31 | 1980 | sock->file->f_flags | file_flags); |
1da177e4 LT |
1981 | out: |
1982 | return err; | |
1983 | } | |
1984 | ||
bd3ded31 JA |
1985 | int __sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
1986 | { | |
1987 | int ret = -EBADF; | |
1988 | struct fd f; | |
1989 | ||
1990 | f = fdget(fd); | |
1991 | if (f.file) { | |
f499a021 JA |
1992 | struct sockaddr_storage address; |
1993 | ||
1994 | ret = move_addr_to_kernel(uservaddr, addrlen, &address); | |
1995 | if (!ret) | |
1996 | ret = __sys_connect_file(f.file, &address, addrlen, 0); | |
6b07edeb | 1997 | fdput(f); |
bd3ded31 JA |
1998 | } |
1999 | ||
2000 | return ret; | |
2001 | } | |
2002 | ||
1387c2c2 DB |
2003 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
2004 | int, addrlen) | |
2005 | { | |
2006 | return __sys_connect(fd, uservaddr, addrlen); | |
2007 | } | |
2008 | ||
1da177e4 LT |
2009 | /* |
2010 | * Get the local address ('name') of a socket object. Move the obtained | |
2011 | * name to user space. | |
2012 | */ | |
2013 | ||
8882a107 DB |
2014 | int __sys_getsockname(int fd, struct sockaddr __user *usockaddr, |
2015 | int __user *usockaddr_len) | |
1da177e4 LT |
2016 | { |
2017 | struct socket *sock; | |
230b1839 | 2018 | struct sockaddr_storage address; |
9b2c45d4 | 2019 | int err, fput_needed; |
89bddce5 | 2020 | |
6cb153ca | 2021 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
2022 | if (!sock) |
2023 | goto out; | |
2024 | ||
2025 | err = security_socket_getsockname(sock); | |
2026 | if (err) | |
2027 | goto out_put; | |
2028 | ||
9b2c45d4 DV |
2029 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 0); |
2030 | if (err < 0) | |
1da177e4 | 2031 | goto out_put; |
e44ef1d4 | 2032 | /* "err" is actually length in this case */ |
9b2c45d4 | 2033 | err = move_addr_to_user(&address, err, usockaddr, usockaddr_len); |
1da177e4 LT |
2034 | |
2035 | out_put: | |
6cb153ca | 2036 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2037 | out: |
2038 | return err; | |
2039 | } | |
2040 | ||
8882a107 DB |
2041 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
2042 | int __user *, usockaddr_len) | |
2043 | { | |
2044 | return __sys_getsockname(fd, usockaddr, usockaddr_len); | |
2045 | } | |
2046 | ||
1da177e4 LT |
2047 | /* |
2048 | * Get the remote address ('name') of a socket object. Move the obtained | |
2049 | * name to user space. | |
2050 | */ | |
2051 | ||
b21c8f83 DB |
2052 | int __sys_getpeername(int fd, struct sockaddr __user *usockaddr, |
2053 | int __user *usockaddr_len) | |
1da177e4 LT |
2054 | { |
2055 | struct socket *sock; | |
230b1839 | 2056 | struct sockaddr_storage address; |
9b2c45d4 | 2057 | int err, fput_needed; |
1da177e4 | 2058 | |
89bddce5 SH |
2059 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2060 | if (sock != NULL) { | |
1da177e4 LT |
2061 | err = security_socket_getpeername(sock); |
2062 | if (err) { | |
6cb153ca | 2063 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2064 | return err; |
2065 | } | |
2066 | ||
9b2c45d4 DV |
2067 | err = sock->ops->getname(sock, (struct sockaddr *)&address, 1); |
2068 | if (err >= 0) | |
2069 | /* "err" is actually length in this case */ | |
2070 | err = move_addr_to_user(&address, err, usockaddr, | |
89bddce5 | 2071 | usockaddr_len); |
6cb153ca | 2072 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2073 | } |
2074 | return err; | |
2075 | } | |
2076 | ||
b21c8f83 DB |
2077 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
2078 | int __user *, usockaddr_len) | |
2079 | { | |
2080 | return __sys_getpeername(fd, usockaddr, usockaddr_len); | |
2081 | } | |
2082 | ||
1da177e4 LT |
2083 | /* |
2084 | * Send a datagram to a given address. We move the address into kernel | |
2085 | * space and check the user space data area is readable before invoking | |
2086 | * the protocol. | |
2087 | */ | |
211b634b DB |
2088 | int __sys_sendto(int fd, void __user *buff, size_t len, unsigned int flags, |
2089 | struct sockaddr __user *addr, int addr_len) | |
1da177e4 LT |
2090 | { |
2091 | struct socket *sock; | |
230b1839 | 2092 | struct sockaddr_storage address; |
1da177e4 LT |
2093 | int err; |
2094 | struct msghdr msg; | |
2095 | struct iovec iov; | |
6cb153ca | 2096 | int fput_needed; |
6cb153ca | 2097 | |
602bd0e9 AV |
2098 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
2099 | if (unlikely(err)) | |
2100 | return err; | |
de0fa95c PE |
2101 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2102 | if (!sock) | |
4387ff75 | 2103 | goto out; |
6cb153ca | 2104 | |
89bddce5 | 2105 | msg.msg_name = NULL; |
89bddce5 SH |
2106 | msg.msg_control = NULL; |
2107 | msg.msg_controllen = 0; | |
2108 | msg.msg_namelen = 0; | |
6cb153ca | 2109 | if (addr) { |
43db362d | 2110 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
2111 | if (err < 0) |
2112 | goto out_put; | |
230b1839 | 2113 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 2114 | msg.msg_namelen = addr_len; |
1da177e4 LT |
2115 | } |
2116 | if (sock->file->f_flags & O_NONBLOCK) | |
2117 | flags |= MSG_DONTWAIT; | |
2118 | msg.msg_flags = flags; | |
d8725c86 | 2119 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 2120 | |
89bddce5 | 2121 | out_put: |
de0fa95c | 2122 | fput_light(sock->file, fput_needed); |
4387ff75 | 2123 | out: |
1da177e4 LT |
2124 | return err; |
2125 | } | |
2126 | ||
211b634b DB |
2127 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
2128 | unsigned int, flags, struct sockaddr __user *, addr, | |
2129 | int, addr_len) | |
2130 | { | |
2131 | return __sys_sendto(fd, buff, len, flags, addr, addr_len); | |
2132 | } | |
2133 | ||
1da177e4 | 2134 | /* |
89bddce5 | 2135 | * Send a datagram down a socket. |
1da177e4 LT |
2136 | */ |
2137 | ||
3e0fa65f | 2138 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 2139 | unsigned int, flags) |
1da177e4 | 2140 | { |
211b634b | 2141 | return __sys_sendto(fd, buff, len, flags, NULL, 0); |
1da177e4 LT |
2142 | } |
2143 | ||
2144 | /* | |
89bddce5 | 2145 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
2146 | * sender. We verify the buffers are writable and if needed move the |
2147 | * sender address from kernel to user space. | |
2148 | */ | |
7a09e1eb DB |
2149 | int __sys_recvfrom(int fd, void __user *ubuf, size_t size, unsigned int flags, |
2150 | struct sockaddr __user *addr, int __user *addr_len) | |
1da177e4 LT |
2151 | { |
2152 | struct socket *sock; | |
2153 | struct iovec iov; | |
2154 | struct msghdr msg; | |
230b1839 | 2155 | struct sockaddr_storage address; |
89bddce5 | 2156 | int err, err2; |
6cb153ca BL |
2157 | int fput_needed; |
2158 | ||
602bd0e9 AV |
2159 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
2160 | if (unlikely(err)) | |
2161 | return err; | |
de0fa95c | 2162 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 2163 | if (!sock) |
de0fa95c | 2164 | goto out; |
1da177e4 | 2165 | |
89bddce5 SH |
2166 | msg.msg_control = NULL; |
2167 | msg.msg_controllen = 0; | |
f3d33426 HFS |
2168 | /* Save some cycles and don't copy the address if not needed */ |
2169 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
2170 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
2171 | msg.msg_namelen = 0; | |
130ed5d1 | 2172 | msg.msg_iocb = NULL; |
9f138fa6 | 2173 | msg.msg_flags = 0; |
1da177e4 LT |
2174 | if (sock->file->f_flags & O_NONBLOCK) |
2175 | flags |= MSG_DONTWAIT; | |
2da62906 | 2176 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 2177 | |
89bddce5 | 2178 | if (err >= 0 && addr != NULL) { |
43db362d | 2179 | err2 = move_addr_to_user(&address, |
230b1839 | 2180 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
2181 | if (err2 < 0) |
2182 | err = err2; | |
1da177e4 | 2183 | } |
de0fa95c PE |
2184 | |
2185 | fput_light(sock->file, fput_needed); | |
4387ff75 | 2186 | out: |
1da177e4 LT |
2187 | return err; |
2188 | } | |
2189 | ||
7a09e1eb DB |
2190 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
2191 | unsigned int, flags, struct sockaddr __user *, addr, | |
2192 | int __user *, addr_len) | |
2193 | { | |
2194 | return __sys_recvfrom(fd, ubuf, size, flags, addr, addr_len); | |
2195 | } | |
2196 | ||
1da177e4 | 2197 | /* |
89bddce5 | 2198 | * Receive a datagram from a socket. |
1da177e4 LT |
2199 | */ |
2200 | ||
b7c0ddf5 JG |
2201 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
2202 | unsigned int, flags) | |
1da177e4 | 2203 | { |
7a09e1eb | 2204 | return __sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
1da177e4 LT |
2205 | } |
2206 | ||
83f0c10b FW |
2207 | static bool sock_use_custom_sol_socket(const struct socket *sock) |
2208 | { | |
2209 | const struct sock *sk = sock->sk; | |
2210 | ||
2211 | /* Use sock->ops->setsockopt() for MPTCP */ | |
2212 | return IS_ENABLED(CONFIG_MPTCP) && | |
2213 | sk->sk_protocol == IPPROTO_MPTCP && | |
2214 | sk->sk_type == SOCK_STREAM && | |
2215 | (sk->sk_family == AF_INET || sk->sk_family == AF_INET6); | |
2216 | } | |
2217 | ||
1da177e4 LT |
2218 | /* |
2219 | * Set a socket option. Because we don't know the option lengths we have | |
2220 | * to pass the user mode parameter for the protocols to sort out. | |
2221 | */ | |
a7b75c5a | 2222 | int __sys_setsockopt(int fd, int level, int optname, char __user *user_optval, |
55db9c0e | 2223 | int optlen) |
1da177e4 | 2224 | { |
519a8a6c | 2225 | sockptr_t optval = USER_SOCKPTR(user_optval); |
0d01da6a | 2226 | char *kernel_optval = NULL; |
6cb153ca | 2227 | int err, fput_needed; |
1da177e4 LT |
2228 | struct socket *sock; |
2229 | ||
2230 | if (optlen < 0) | |
2231 | return -EINVAL; | |
89bddce5 SH |
2232 | |
2233 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
4a367299 CH |
2234 | if (!sock) |
2235 | return err; | |
1da177e4 | 2236 | |
4a367299 CH |
2237 | err = security_socket_setsockopt(sock, level, optname); |
2238 | if (err) | |
2239 | goto out_put; | |
0d01da6a | 2240 | |
55db9c0e CH |
2241 | if (!in_compat_syscall()) |
2242 | err = BPF_CGROUP_RUN_PROG_SETSOCKOPT(sock->sk, &level, &optname, | |
a7b75c5a | 2243 | user_optval, &optlen, |
55db9c0e | 2244 | &kernel_optval); |
4a367299 CH |
2245 | if (err < 0) |
2246 | goto out_put; | |
2247 | if (err > 0) { | |
2248 | err = 0; | |
2249 | goto out_put; | |
2250 | } | |
0d01da6a | 2251 | |
a7b75c5a CH |
2252 | if (kernel_optval) |
2253 | optval = KERNEL_SOCKPTR(kernel_optval); | |
4a367299 | 2254 | if (level == SOL_SOCKET && !sock_use_custom_sol_socket(sock)) |
a7b75c5a | 2255 | err = sock_setsockopt(sock, level, optname, optval, optlen); |
a44d9e72 CH |
2256 | else if (unlikely(!sock->ops->setsockopt)) |
2257 | err = -EOPNOTSUPP; | |
4a367299 CH |
2258 | else |
2259 | err = sock->ops->setsockopt(sock, level, optname, optval, | |
89bddce5 | 2260 | optlen); |
a7b75c5a | 2261 | kfree(kernel_optval); |
4a367299 CH |
2262 | out_put: |
2263 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
2264 | return err; |
2265 | } | |
2266 | ||
cc36dca0 DB |
2267 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
2268 | char __user *, optval, int, optlen) | |
2269 | { | |
2270 | return __sys_setsockopt(fd, level, optname, optval, optlen); | |
2271 | } | |
2272 | ||
9cacf81f SF |
2273 | INDIRECT_CALLABLE_DECLARE(bool tcp_bpf_bypass_getsockopt(int level, |
2274 | int optname)); | |
2275 | ||
1da177e4 LT |
2276 | /* |
2277 | * Get a socket option. Because we don't know the option lengths we have | |
2278 | * to pass a user mode parameter for the protocols to sort out. | |
2279 | */ | |
55db9c0e CH |
2280 | int __sys_getsockopt(int fd, int level, int optname, char __user *optval, |
2281 | int __user *optlen) | |
1da177e4 | 2282 | { |
6cb153ca | 2283 | int err, fput_needed; |
1da177e4 | 2284 | struct socket *sock; |
0d01da6a | 2285 | int max_optlen; |
1da177e4 | 2286 | |
89bddce5 | 2287 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
d8a9b38f CH |
2288 | if (!sock) |
2289 | return err; | |
2290 | ||
2291 | err = security_socket_getsockopt(sock, level, optname); | |
2292 | if (err) | |
2293 | goto out_put; | |
1da177e4 | 2294 | |
55db9c0e CH |
2295 | if (!in_compat_syscall()) |
2296 | max_optlen = BPF_CGROUP_GETSOCKOPT_MAX_OPTLEN(optlen); | |
0d01da6a | 2297 | |
d8a9b38f CH |
2298 | if (level == SOL_SOCKET) |
2299 | err = sock_getsockopt(sock, level, optname, optval, optlen); | |
a44d9e72 CH |
2300 | else if (unlikely(!sock->ops->getsockopt)) |
2301 | err = -EOPNOTSUPP; | |
d8a9b38f CH |
2302 | else |
2303 | err = sock->ops->getsockopt(sock, level, optname, optval, | |
89bddce5 | 2304 | optlen); |
0d01da6a | 2305 | |
55db9c0e CH |
2306 | if (!in_compat_syscall()) |
2307 | err = BPF_CGROUP_RUN_PROG_GETSOCKOPT(sock->sk, level, optname, | |
2308 | optval, optlen, max_optlen, | |
2309 | err); | |
6cb153ca | 2310 | out_put: |
d8a9b38f | 2311 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2312 | return err; |
2313 | } | |
2314 | ||
13a2d70e DB |
2315 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
2316 | char __user *, optval, int __user *, optlen) | |
2317 | { | |
2318 | return __sys_getsockopt(fd, level, optname, optval, optlen); | |
2319 | } | |
2320 | ||
1da177e4 LT |
2321 | /* |
2322 | * Shutdown a socket. | |
2323 | */ | |
2324 | ||
b713c195 JA |
2325 | int __sys_shutdown_sock(struct socket *sock, int how) |
2326 | { | |
2327 | int err; | |
2328 | ||
2329 | err = security_socket_shutdown(sock, how); | |
2330 | if (!err) | |
2331 | err = sock->ops->shutdown(sock, how); | |
2332 | ||
2333 | return err; | |
2334 | } | |
2335 | ||
005a1aea | 2336 | int __sys_shutdown(int fd, int how) |
1da177e4 | 2337 | { |
6cb153ca | 2338 | int err, fput_needed; |
1da177e4 LT |
2339 | struct socket *sock; |
2340 | ||
89bddce5 SH |
2341 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
2342 | if (sock != NULL) { | |
b713c195 | 2343 | err = __sys_shutdown_sock(sock, how); |
6cb153ca | 2344 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2345 | } |
2346 | return err; | |
2347 | } | |
2348 | ||
005a1aea DB |
2349 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
2350 | { | |
2351 | return __sys_shutdown(fd, how); | |
2352 | } | |
2353 | ||
89bddce5 | 2354 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
2355 | * fields which are the same type (int / unsigned) on our platforms. |
2356 | */ | |
2357 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
2358 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
2359 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
2360 | ||
c71d8ebe TH |
2361 | struct used_address { |
2362 | struct sockaddr_storage name; | |
2363 | unsigned int name_len; | |
2364 | }; | |
2365 | ||
0a384abf JA |
2366 | int __copy_msghdr_from_user(struct msghdr *kmsg, |
2367 | struct user_msghdr __user *umsg, | |
2368 | struct sockaddr __user **save_addr, | |
2369 | struct iovec __user **uiov, size_t *nsegs) | |
1661bf36 | 2370 | { |
ffb07550 | 2371 | struct user_msghdr msg; |
08adb7da AV |
2372 | ssize_t err; |
2373 | ||
ffb07550 | 2374 | if (copy_from_user(&msg, umsg, sizeof(*umsg))) |
1661bf36 | 2375 | return -EFAULT; |
dbb490b9 | 2376 | |
1f466e1f CH |
2377 | kmsg->msg_control_is_user = true; |
2378 | kmsg->msg_control_user = msg.msg_control; | |
ffb07550 AV |
2379 | kmsg->msg_controllen = msg.msg_controllen; |
2380 | kmsg->msg_flags = msg.msg_flags; | |
2381 | ||
2382 | kmsg->msg_namelen = msg.msg_namelen; | |
2383 | if (!msg.msg_name) | |
6a2a2b3a AS |
2384 | kmsg->msg_namelen = 0; |
2385 | ||
dbb490b9 ML |
2386 | if (kmsg->msg_namelen < 0) |
2387 | return -EINVAL; | |
2388 | ||
1661bf36 | 2389 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 2390 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
2391 | |
2392 | if (save_addr) | |
ffb07550 | 2393 | *save_addr = msg.msg_name; |
08adb7da | 2394 | |
ffb07550 | 2395 | if (msg.msg_name && kmsg->msg_namelen) { |
08adb7da | 2396 | if (!save_addr) { |
864d9664 PA |
2397 | err = move_addr_to_kernel(msg.msg_name, |
2398 | kmsg->msg_namelen, | |
08adb7da AV |
2399 | kmsg->msg_name); |
2400 | if (err < 0) | |
2401 | return err; | |
2402 | } | |
2403 | } else { | |
2404 | kmsg->msg_name = NULL; | |
2405 | kmsg->msg_namelen = 0; | |
2406 | } | |
2407 | ||
ffb07550 | 2408 | if (msg.msg_iovlen > UIO_MAXIOV) |
08adb7da AV |
2409 | return -EMSGSIZE; |
2410 | ||
0345f931 | 2411 | kmsg->msg_iocb = NULL; |
0a384abf JA |
2412 | *uiov = msg.msg_iov; |
2413 | *nsegs = msg.msg_iovlen; | |
2414 | return 0; | |
2415 | } | |
2416 | ||
2417 | static int copy_msghdr_from_user(struct msghdr *kmsg, | |
2418 | struct user_msghdr __user *umsg, | |
2419 | struct sockaddr __user **save_addr, | |
2420 | struct iovec **iov) | |
2421 | { | |
2422 | struct user_msghdr msg; | |
2423 | ssize_t err; | |
2424 | ||
2425 | err = __copy_msghdr_from_user(kmsg, umsg, save_addr, &msg.msg_iov, | |
2426 | &msg.msg_iovlen); | |
2427 | if (err) | |
2428 | return err; | |
0345f931 | 2429 | |
87e5e6da | 2430 | err = import_iovec(save_addr ? READ : WRITE, |
ffb07550 | 2431 | msg.msg_iov, msg.msg_iovlen, |
da184284 | 2432 | UIO_FASTIOV, iov, &kmsg->msg_iter); |
87e5e6da | 2433 | return err < 0 ? err : 0; |
1661bf36 DC |
2434 | } |
2435 | ||
4257c8ca JA |
2436 | static int ____sys_sendmsg(struct socket *sock, struct msghdr *msg_sys, |
2437 | unsigned int flags, struct used_address *used_address, | |
2438 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 2439 | { |
b9d717a7 | 2440 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
846cc123 | 2441 | __aligned(sizeof(__kernel_size_t)); |
89bddce5 | 2442 | /* 20 is size of ipv6_pktinfo */ |
1da177e4 | 2443 | unsigned char *ctl_buf = ctl; |
d8725c86 | 2444 | int ctl_len; |
08adb7da | 2445 | ssize_t err; |
89bddce5 | 2446 | |
1da177e4 LT |
2447 | err = -ENOBUFS; |
2448 | ||
228e548e | 2449 | if (msg_sys->msg_controllen > INT_MAX) |
4257c8ca | 2450 | goto out; |
28a94d8f | 2451 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 2452 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2453 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2454 | err = |
228e548e | 2455 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2456 | sizeof(ctl)); |
1da177e4 | 2457 | if (err) |
4257c8ca | 2458 | goto out; |
228e548e AB |
2459 | ctl_buf = msg_sys->msg_control; |
2460 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2461 | } else if (ctl_len) { |
ac4340fc DM |
2462 | BUILD_BUG_ON(sizeof(struct cmsghdr) != |
2463 | CMSG_ALIGN(sizeof(struct cmsghdr))); | |
89bddce5 | 2464 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2465 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2466 | if (ctl_buf == NULL) |
4257c8ca | 2467 | goto out; |
1da177e4 LT |
2468 | } |
2469 | err = -EFAULT; | |
1f466e1f | 2470 | if (copy_from_user(ctl_buf, msg_sys->msg_control_user, ctl_len)) |
1da177e4 | 2471 | goto out_freectl; |
228e548e | 2472 | msg_sys->msg_control = ctl_buf; |
1f466e1f | 2473 | msg_sys->msg_control_is_user = false; |
1da177e4 | 2474 | } |
228e548e | 2475 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2476 | |
2477 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2478 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2479 | /* |
2480 | * If this is sendmmsg() and current destination address is same as | |
2481 | * previously succeeded address, omit asking LSM's decision. | |
2482 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2483 | * destination address never matches. | |
2484 | */ | |
bc909d9d MD |
2485 | if (used_address && msg_sys->msg_name && |
2486 | used_address->name_len == msg_sys->msg_namelen && | |
2487 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 2488 | used_address->name_len)) { |
d8725c86 | 2489 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
2490 | goto out_freectl; |
2491 | } | |
d8725c86 | 2492 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
2493 | /* |
2494 | * If this is sendmmsg() and sending to current destination address was | |
2495 | * successful, remember it. | |
2496 | */ | |
2497 | if (used_address && err >= 0) { | |
2498 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2499 | if (msg_sys->msg_name) |
2500 | memcpy(&used_address->name, msg_sys->msg_name, | |
2501 | used_address->name_len); | |
c71d8ebe | 2502 | } |
1da177e4 LT |
2503 | |
2504 | out_freectl: | |
89bddce5 | 2505 | if (ctl_buf != ctl) |
1da177e4 | 2506 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
4257c8ca JA |
2507 | out: |
2508 | return err; | |
2509 | } | |
2510 | ||
03b1230c JA |
2511 | int sendmsg_copy_msghdr(struct msghdr *msg, |
2512 | struct user_msghdr __user *umsg, unsigned flags, | |
2513 | struct iovec **iov) | |
4257c8ca JA |
2514 | { |
2515 | int err; | |
2516 | ||
2517 | if (flags & MSG_CMSG_COMPAT) { | |
2518 | struct compat_msghdr __user *msg_compat; | |
2519 | ||
2520 | msg_compat = (struct compat_msghdr __user *) umsg; | |
2521 | err = get_compat_msghdr(msg, msg_compat, NULL, iov); | |
2522 | } else { | |
2523 | err = copy_msghdr_from_user(msg, umsg, NULL, iov); | |
2524 | } | |
2525 | if (err < 0) | |
2526 | return err; | |
2527 | ||
2528 | return 0; | |
2529 | } | |
2530 | ||
2531 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, | |
2532 | struct msghdr *msg_sys, unsigned int flags, | |
2533 | struct used_address *used_address, | |
2534 | unsigned int allowed_msghdr_flags) | |
2535 | { | |
2536 | struct sockaddr_storage address; | |
2537 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; | |
2538 | ssize_t err; | |
2539 | ||
2540 | msg_sys->msg_name = &address; | |
2541 | ||
2542 | err = sendmsg_copy_msghdr(msg_sys, msg, flags, &iov); | |
2543 | if (err < 0) | |
2544 | return err; | |
2545 | ||
2546 | err = ____sys_sendmsg(sock, msg_sys, flags, used_address, | |
2547 | allowed_msghdr_flags); | |
da184284 | 2548 | kfree(iov); |
228e548e AB |
2549 | return err; |
2550 | } | |
2551 | ||
2552 | /* | |
2553 | * BSD sendmsg interface | |
2554 | */ | |
03b1230c | 2555 | long __sys_sendmsg_sock(struct socket *sock, struct msghdr *msg, |
0fa03c62 JA |
2556 | unsigned int flags) |
2557 | { | |
03b1230c | 2558 | return ____sys_sendmsg(sock, msg, flags, NULL, 0); |
0fa03c62 | 2559 | } |
228e548e | 2560 | |
e1834a32 DB |
2561 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2562 | bool forbid_cmsg_compat) | |
228e548e AB |
2563 | { |
2564 | int fput_needed, err; | |
2565 | struct msghdr msg_sys; | |
1be374a0 AL |
2566 | struct socket *sock; |
2567 | ||
e1834a32 DB |
2568 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2569 | return -EINVAL; | |
2570 | ||
1be374a0 | 2571 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2572 | if (!sock) |
2573 | goto out; | |
2574 | ||
28a94d8f | 2575 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 2576 | |
6cb153ca | 2577 | fput_light(sock->file, fput_needed); |
89bddce5 | 2578 | out: |
1da177e4 LT |
2579 | return err; |
2580 | } | |
2581 | ||
666547ff | 2582 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 | 2583 | { |
e1834a32 | 2584 | return __sys_sendmsg(fd, msg, flags, true); |
a7526eb5 AL |
2585 | } |
2586 | ||
228e548e AB |
2587 | /* |
2588 | * Linux sendmmsg interface | |
2589 | */ | |
2590 | ||
2591 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
e1834a32 | 2592 | unsigned int flags, bool forbid_cmsg_compat) |
228e548e AB |
2593 | { |
2594 | int fput_needed, err, datagrams; | |
2595 | struct socket *sock; | |
2596 | struct mmsghdr __user *entry; | |
2597 | struct compat_mmsghdr __user *compat_entry; | |
2598 | struct msghdr msg_sys; | |
c71d8ebe | 2599 | struct used_address used_address; |
f092276d | 2600 | unsigned int oflags = flags; |
228e548e | 2601 | |
e1834a32 DB |
2602 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2603 | return -EINVAL; | |
2604 | ||
98382f41 AB |
2605 | if (vlen > UIO_MAXIOV) |
2606 | vlen = UIO_MAXIOV; | |
228e548e AB |
2607 | |
2608 | datagrams = 0; | |
2609 | ||
2610 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2611 | if (!sock) | |
2612 | return err; | |
2613 | ||
c71d8ebe | 2614 | used_address.name_len = UINT_MAX; |
228e548e AB |
2615 | entry = mmsg; |
2616 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2617 | err = 0; |
f092276d | 2618 | flags |= MSG_BATCH; |
228e548e AB |
2619 | |
2620 | while (datagrams < vlen) { | |
f092276d TH |
2621 | if (datagrams == vlen - 1) |
2622 | flags = oflags; | |
2623 | ||
228e548e | 2624 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2625 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2626 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2627 | if (err < 0) |
2628 | break; | |
2629 | err = __put_user(err, &compat_entry->msg_len); | |
2630 | ++compat_entry; | |
2631 | } else { | |
a7526eb5 | 2632 | err = ___sys_sendmsg(sock, |
666547ff | 2633 | (struct user_msghdr __user *)entry, |
28a94d8f | 2634 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2635 | if (err < 0) |
2636 | break; | |
2637 | err = put_user(err, &entry->msg_len); | |
2638 | ++entry; | |
2639 | } | |
2640 | ||
2641 | if (err) | |
2642 | break; | |
2643 | ++datagrams; | |
3023898b SHY |
2644 | if (msg_data_left(&msg_sys)) |
2645 | break; | |
a78cb84c | 2646 | cond_resched(); |
228e548e AB |
2647 | } |
2648 | ||
228e548e AB |
2649 | fput_light(sock->file, fput_needed); |
2650 | ||
728ffb86 AB |
2651 | /* We only return an error if no datagrams were able to be sent */ |
2652 | if (datagrams != 0) | |
228e548e AB |
2653 | return datagrams; |
2654 | ||
228e548e AB |
2655 | return err; |
2656 | } | |
2657 | ||
2658 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2659 | unsigned int, vlen, unsigned int, flags) | |
2660 | { | |
e1834a32 | 2661 | return __sys_sendmmsg(fd, mmsg, vlen, flags, true); |
228e548e AB |
2662 | } |
2663 | ||
03b1230c JA |
2664 | int recvmsg_copy_msghdr(struct msghdr *msg, |
2665 | struct user_msghdr __user *umsg, unsigned flags, | |
2666 | struct sockaddr __user **uaddr, | |
2667 | struct iovec **iov) | |
1da177e4 | 2668 | { |
08adb7da | 2669 | ssize_t err; |
1da177e4 | 2670 | |
4257c8ca JA |
2671 | if (MSG_CMSG_COMPAT & flags) { |
2672 | struct compat_msghdr __user *msg_compat; | |
1da177e4 | 2673 | |
4257c8ca JA |
2674 | msg_compat = (struct compat_msghdr __user *) umsg; |
2675 | err = get_compat_msghdr(msg, msg_compat, uaddr, iov); | |
2676 | } else { | |
2677 | err = copy_msghdr_from_user(msg, umsg, uaddr, iov); | |
2678 | } | |
1da177e4 | 2679 | if (err < 0) |
da184284 | 2680 | return err; |
1da177e4 | 2681 | |
4257c8ca JA |
2682 | return 0; |
2683 | } | |
2684 | ||
2685 | static int ____sys_recvmsg(struct socket *sock, struct msghdr *msg_sys, | |
2686 | struct user_msghdr __user *msg, | |
2687 | struct sockaddr __user *uaddr, | |
2688 | unsigned int flags, int nosec) | |
2689 | { | |
2690 | struct compat_msghdr __user *msg_compat = | |
2691 | (struct compat_msghdr __user *) msg; | |
2692 | int __user *uaddr_len = COMPAT_NAMELEN(msg); | |
2693 | struct sockaddr_storage addr; | |
2694 | unsigned long cmsg_ptr; | |
2695 | int len; | |
2696 | ssize_t err; | |
2697 | ||
2698 | msg_sys->msg_name = &addr; | |
a2e27255 ACM |
2699 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2700 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2701 | |
f3d33426 HFS |
2702 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2703 | msg_sys->msg_namelen = 0; | |
2704 | ||
1da177e4 LT |
2705 | if (sock->file->f_flags & O_NONBLOCK) |
2706 | flags |= MSG_DONTWAIT; | |
1af66221 ED |
2707 | |
2708 | if (unlikely(nosec)) | |
2709 | err = sock_recvmsg_nosec(sock, msg_sys, flags); | |
2710 | else | |
2711 | err = sock_recvmsg(sock, msg_sys, flags); | |
2712 | ||
1da177e4 | 2713 | if (err < 0) |
4257c8ca | 2714 | goto out; |
1da177e4 LT |
2715 | len = err; |
2716 | ||
2717 | if (uaddr != NULL) { | |
43db362d | 2718 | err = move_addr_to_user(&addr, |
a2e27255 | 2719 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2720 | uaddr_len); |
1da177e4 | 2721 | if (err < 0) |
4257c8ca | 2722 | goto out; |
1da177e4 | 2723 | } |
a2e27255 | 2724 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2725 | COMPAT_FLAGS(msg)); |
1da177e4 | 2726 | if (err) |
4257c8ca | 2727 | goto out; |
1da177e4 | 2728 | if (MSG_CMSG_COMPAT & flags) |
a2e27255 | 2729 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2730 | &msg_compat->msg_controllen); |
2731 | else | |
a2e27255 | 2732 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2733 | &msg->msg_controllen); |
2734 | if (err) | |
4257c8ca | 2735 | goto out; |
1da177e4 | 2736 | err = len; |
4257c8ca JA |
2737 | out: |
2738 | return err; | |
2739 | } | |
2740 | ||
2741 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, | |
2742 | struct msghdr *msg_sys, unsigned int flags, int nosec) | |
2743 | { | |
2744 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; | |
2745 | /* user mode address pointers */ | |
2746 | struct sockaddr __user *uaddr; | |
2747 | ssize_t err; | |
2748 | ||
2749 | err = recvmsg_copy_msghdr(msg_sys, msg, flags, &uaddr, &iov); | |
2750 | if (err < 0) | |
2751 | return err; | |
1da177e4 | 2752 | |
4257c8ca | 2753 | err = ____sys_recvmsg(sock, msg_sys, msg, uaddr, flags, nosec); |
da184284 | 2754 | kfree(iov); |
a2e27255 ACM |
2755 | return err; |
2756 | } | |
2757 | ||
2758 | /* | |
2759 | * BSD recvmsg interface | |
2760 | */ | |
2761 | ||
03b1230c JA |
2762 | long __sys_recvmsg_sock(struct socket *sock, struct msghdr *msg, |
2763 | struct user_msghdr __user *umsg, | |
2764 | struct sockaddr __user *uaddr, unsigned int flags) | |
aa1fa28f | 2765 | { |
03b1230c | 2766 | return ____sys_recvmsg(sock, msg, umsg, uaddr, flags, 0); |
aa1fa28f JA |
2767 | } |
2768 | ||
e1834a32 DB |
2769 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned int flags, |
2770 | bool forbid_cmsg_compat) | |
a2e27255 ACM |
2771 | { |
2772 | int fput_needed, err; | |
2773 | struct msghdr msg_sys; | |
1be374a0 AL |
2774 | struct socket *sock; |
2775 | ||
e1834a32 DB |
2776 | if (forbid_cmsg_compat && (flags & MSG_CMSG_COMPAT)) |
2777 | return -EINVAL; | |
2778 | ||
1be374a0 | 2779 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2780 | if (!sock) |
2781 | goto out; | |
2782 | ||
a7526eb5 | 2783 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2784 | |
6cb153ca | 2785 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2786 | out: |
2787 | return err; | |
2788 | } | |
2789 | ||
666547ff | 2790 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2791 | unsigned int, flags) |
2792 | { | |
e1834a32 | 2793 | return __sys_recvmsg(fd, msg, flags, true); |
a7526eb5 AL |
2794 | } |
2795 | ||
a2e27255 ACM |
2796 | /* |
2797 | * Linux recvmmsg interface | |
2798 | */ | |
2799 | ||
e11d4284 AB |
2800 | static int do_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2801 | unsigned int vlen, unsigned int flags, | |
2802 | struct timespec64 *timeout) | |
a2e27255 ACM |
2803 | { |
2804 | int fput_needed, err, datagrams; | |
2805 | struct socket *sock; | |
2806 | struct mmsghdr __user *entry; | |
d7256d0e | 2807 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2808 | struct msghdr msg_sys; |
766b9f92 DD |
2809 | struct timespec64 end_time; |
2810 | struct timespec64 timeout64; | |
a2e27255 ACM |
2811 | |
2812 | if (timeout && | |
2813 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2814 | timeout->tv_nsec)) | |
2815 | return -EINVAL; | |
2816 | ||
2817 | datagrams = 0; | |
2818 | ||
2819 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2820 | if (!sock) | |
2821 | return err; | |
2822 | ||
7797dc41 SHY |
2823 | if (likely(!(flags & MSG_ERRQUEUE))) { |
2824 | err = sock_error(sock->sk); | |
2825 | if (err) { | |
2826 | datagrams = err; | |
2827 | goto out_put; | |
2828 | } | |
e623a9e9 | 2829 | } |
a2e27255 ACM |
2830 | |
2831 | entry = mmsg; | |
d7256d0e | 2832 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2833 | |
2834 | while (datagrams < vlen) { | |
2835 | /* | |
2836 | * No need to ask LSM for more than the first datagram. | |
2837 | */ | |
d7256d0e | 2838 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2839 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2840 | &msg_sys, flags & ~MSG_WAITFORONE, |
2841 | datagrams); | |
d7256d0e JMG |
2842 | if (err < 0) |
2843 | break; | |
2844 | err = __put_user(err, &compat_entry->msg_len); | |
2845 | ++compat_entry; | |
2846 | } else { | |
a7526eb5 | 2847 | err = ___sys_recvmsg(sock, |
666547ff | 2848 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2849 | &msg_sys, flags & ~MSG_WAITFORONE, |
2850 | datagrams); | |
d7256d0e JMG |
2851 | if (err < 0) |
2852 | break; | |
2853 | err = put_user(err, &entry->msg_len); | |
2854 | ++entry; | |
2855 | } | |
2856 | ||
a2e27255 ACM |
2857 | if (err) |
2858 | break; | |
a2e27255 ACM |
2859 | ++datagrams; |
2860 | ||
71c5c159 BB |
2861 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2862 | if (flags & MSG_WAITFORONE) | |
2863 | flags |= MSG_DONTWAIT; | |
2864 | ||
a2e27255 | 2865 | if (timeout) { |
766b9f92 | 2866 | ktime_get_ts64(&timeout64); |
c2e6c856 | 2867 | *timeout = timespec64_sub(end_time, timeout64); |
a2e27255 ACM |
2868 | if (timeout->tv_sec < 0) { |
2869 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2870 | break; | |
2871 | } | |
2872 | ||
2873 | /* Timeout, return less than vlen datagrams */ | |
2874 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2875 | break; | |
2876 | } | |
2877 | ||
2878 | /* Out of band data, return right away */ | |
2879 | if (msg_sys.msg_flags & MSG_OOB) | |
2880 | break; | |
a78cb84c | 2881 | cond_resched(); |
a2e27255 ACM |
2882 | } |
2883 | ||
a2e27255 | 2884 | if (err == 0) |
34b88a68 ACM |
2885 | goto out_put; |
2886 | ||
2887 | if (datagrams == 0) { | |
2888 | datagrams = err; | |
2889 | goto out_put; | |
2890 | } | |
a2e27255 | 2891 | |
34b88a68 ACM |
2892 | /* |
2893 | * We may return less entries than requested (vlen) if the | |
2894 | * sock is non block and there aren't enough datagrams... | |
2895 | */ | |
2896 | if (err != -EAGAIN) { | |
a2e27255 | 2897 | /* |
34b88a68 ACM |
2898 | * ... or if recvmsg returns an error after we |
2899 | * received some datagrams, where we record the | |
2900 | * error to return on the next call or if the | |
2901 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2902 | */ |
34b88a68 | 2903 | sock->sk->sk_err = -err; |
a2e27255 | 2904 | } |
34b88a68 ACM |
2905 | out_put: |
2906 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2907 | |
34b88a68 | 2908 | return datagrams; |
a2e27255 ACM |
2909 | } |
2910 | ||
e11d4284 AB |
2911 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, |
2912 | unsigned int vlen, unsigned int flags, | |
2913 | struct __kernel_timespec __user *timeout, | |
2914 | struct old_timespec32 __user *timeout32) | |
a2e27255 ACM |
2915 | { |
2916 | int datagrams; | |
c2e6c856 | 2917 | struct timespec64 timeout_sys; |
a2e27255 | 2918 | |
e11d4284 AB |
2919 | if (timeout && get_timespec64(&timeout_sys, timeout)) |
2920 | return -EFAULT; | |
a2e27255 | 2921 | |
e11d4284 | 2922 | if (timeout32 && get_old_timespec32(&timeout_sys, timeout32)) |
a2e27255 ACM |
2923 | return -EFAULT; |
2924 | ||
e11d4284 AB |
2925 | if (!timeout && !timeout32) |
2926 | return do_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2927 | ||
2928 | datagrams = do_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
a2e27255 | 2929 | |
e11d4284 AB |
2930 | if (datagrams <= 0) |
2931 | return datagrams; | |
2932 | ||
2933 | if (timeout && put_timespec64(&timeout_sys, timeout)) | |
2934 | datagrams = -EFAULT; | |
2935 | ||
2936 | if (timeout32 && put_old_timespec32(&timeout_sys, timeout32)) | |
a2e27255 ACM |
2937 | datagrams = -EFAULT; |
2938 | ||
2939 | return datagrams; | |
2940 | } | |
2941 | ||
1255e269 DB |
2942 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, |
2943 | unsigned int, vlen, unsigned int, flags, | |
c2e6c856 | 2944 | struct __kernel_timespec __user *, timeout) |
1255e269 | 2945 | { |
e11d4284 AB |
2946 | if (flags & MSG_CMSG_COMPAT) |
2947 | return -EINVAL; | |
2948 | ||
2949 | return __sys_recvmmsg(fd, mmsg, vlen, flags, timeout, NULL); | |
2950 | } | |
2951 | ||
2952 | #ifdef CONFIG_COMPAT_32BIT_TIME | |
2953 | SYSCALL_DEFINE5(recvmmsg_time32, int, fd, struct mmsghdr __user *, mmsg, | |
2954 | unsigned int, vlen, unsigned int, flags, | |
2955 | struct old_timespec32 __user *, timeout) | |
2956 | { | |
2957 | if (flags & MSG_CMSG_COMPAT) | |
2958 | return -EINVAL; | |
2959 | ||
2960 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL, timeout); | |
1255e269 | 2961 | } |
e11d4284 | 2962 | #endif |
1255e269 | 2963 | |
a2e27255 | 2964 | #ifdef __ARCH_WANT_SYS_SOCKETCALL |
1da177e4 LT |
2965 | /* Argument list sizes for sys_socketcall */ |
2966 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2967 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2968 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2969 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2970 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2971 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2972 | }; |
2973 | ||
1da177e4 LT |
2974 | #undef AL |
2975 | ||
2976 | /* | |
89bddce5 | 2977 | * System call vectors. |
1da177e4 LT |
2978 | * |
2979 | * Argument checking cleaned up. Saved 20% in size. | |
2980 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2981 | * it is set by the callees. |
1da177e4 LT |
2982 | */ |
2983 | ||
3e0fa65f | 2984 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2985 | { |
2950fa9d | 2986 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2987 | unsigned long a0, a1; |
1da177e4 | 2988 | int err; |
47379052 | 2989 | unsigned int len; |
1da177e4 | 2990 | |
228e548e | 2991 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 | 2992 | return -EINVAL; |
c8e8cd57 | 2993 | call = array_index_nospec(call, SYS_SENDMMSG + 1); |
1da177e4 | 2994 | |
47379052 AV |
2995 | len = nargs[call]; |
2996 | if (len > sizeof(a)) | |
2997 | return -EINVAL; | |
2998 | ||
1da177e4 | 2999 | /* copy_from_user should be SMP safe. */ |
47379052 | 3000 | if (copy_from_user(a, args, len)) |
1da177e4 | 3001 | return -EFAULT; |
3ec3b2fb | 3002 | |
2950fa9d CG |
3003 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
3004 | if (err) | |
3005 | return err; | |
3ec3b2fb | 3006 | |
89bddce5 SH |
3007 | a0 = a[0]; |
3008 | a1 = a[1]; | |
3009 | ||
3010 | switch (call) { | |
3011 | case SYS_SOCKET: | |
9d6a15c3 | 3012 | err = __sys_socket(a0, a1, a[2]); |
89bddce5 SH |
3013 | break; |
3014 | case SYS_BIND: | |
a87d35d8 | 3015 | err = __sys_bind(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
3016 | break; |
3017 | case SYS_CONNECT: | |
1387c2c2 | 3018 | err = __sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
89bddce5 SH |
3019 | break; |
3020 | case SYS_LISTEN: | |
25e290ee | 3021 | err = __sys_listen(a0, a1); |
89bddce5 SH |
3022 | break; |
3023 | case SYS_ACCEPT: | |
4541e805 DB |
3024 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
3025 | (int __user *)a[2], 0); | |
89bddce5 SH |
3026 | break; |
3027 | case SYS_GETSOCKNAME: | |
3028 | err = | |
8882a107 DB |
3029 | __sys_getsockname(a0, (struct sockaddr __user *)a1, |
3030 | (int __user *)a[2]); | |
89bddce5 SH |
3031 | break; |
3032 | case SYS_GETPEERNAME: | |
3033 | err = | |
b21c8f83 DB |
3034 | __sys_getpeername(a0, (struct sockaddr __user *)a1, |
3035 | (int __user *)a[2]); | |
89bddce5 SH |
3036 | break; |
3037 | case SYS_SOCKETPAIR: | |
6debc8d8 | 3038 | err = __sys_socketpair(a0, a1, a[2], (int __user *)a[3]); |
89bddce5 SH |
3039 | break; |
3040 | case SYS_SEND: | |
f3bf896b DB |
3041 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
3042 | NULL, 0); | |
89bddce5 SH |
3043 | break; |
3044 | case SYS_SENDTO: | |
211b634b DB |
3045 | err = __sys_sendto(a0, (void __user *)a1, a[2], a[3], |
3046 | (struct sockaddr __user *)a[4], a[5]); | |
89bddce5 SH |
3047 | break; |
3048 | case SYS_RECV: | |
d27e9afc DB |
3049 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
3050 | NULL, NULL); | |
89bddce5 SH |
3051 | break; |
3052 | case SYS_RECVFROM: | |
7a09e1eb DB |
3053 | err = __sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
3054 | (struct sockaddr __user *)a[4], | |
3055 | (int __user *)a[5]); | |
89bddce5 SH |
3056 | break; |
3057 | case SYS_SHUTDOWN: | |
005a1aea | 3058 | err = __sys_shutdown(a0, a1); |
89bddce5 SH |
3059 | break; |
3060 | case SYS_SETSOCKOPT: | |
cc36dca0 DB |
3061 | err = __sys_setsockopt(a0, a1, a[2], (char __user *)a[3], |
3062 | a[4]); | |
89bddce5 SH |
3063 | break; |
3064 | case SYS_GETSOCKOPT: | |
3065 | err = | |
13a2d70e DB |
3066 | __sys_getsockopt(a0, a1, a[2], (char __user *)a[3], |
3067 | (int __user *)a[4]); | |
89bddce5 SH |
3068 | break; |
3069 | case SYS_SENDMSG: | |
e1834a32 DB |
3070 | err = __sys_sendmsg(a0, (struct user_msghdr __user *)a1, |
3071 | a[2], true); | |
89bddce5 | 3072 | break; |
228e548e | 3073 | case SYS_SENDMMSG: |
e1834a32 DB |
3074 | err = __sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], |
3075 | a[3], true); | |
228e548e | 3076 | break; |
89bddce5 | 3077 | case SYS_RECVMSG: |
e1834a32 DB |
3078 | err = __sys_recvmsg(a0, (struct user_msghdr __user *)a1, |
3079 | a[2], true); | |
89bddce5 | 3080 | break; |
a2e27255 | 3081 | case SYS_RECVMMSG: |
3ca47e95 | 3082 | if (IS_ENABLED(CONFIG_64BIT)) |
e11d4284 AB |
3083 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, |
3084 | a[2], a[3], | |
3085 | (struct __kernel_timespec __user *)a[4], | |
3086 | NULL); | |
3087 | else | |
3088 | err = __sys_recvmmsg(a0, (struct mmsghdr __user *)a1, | |
3089 | a[2], a[3], NULL, | |
3090 | (struct old_timespec32 __user *)a[4]); | |
a2e27255 | 3091 | break; |
de11defe | 3092 | case SYS_ACCEPT4: |
4541e805 DB |
3093 | err = __sys_accept4(a0, (struct sockaddr __user *)a1, |
3094 | (int __user *)a[2], a[3]); | |
aaca0bdc | 3095 | break; |
89bddce5 SH |
3096 | default: |
3097 | err = -EINVAL; | |
3098 | break; | |
1da177e4 LT |
3099 | } |
3100 | return err; | |
3101 | } | |
3102 | ||
89bddce5 | 3103 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 3104 | |
55737fda SH |
3105 | /** |
3106 | * sock_register - add a socket protocol handler | |
3107 | * @ops: description of protocol | |
3108 | * | |
1da177e4 LT |
3109 | * This function is called by a protocol handler that wants to |
3110 | * advertise its address family, and have it linked into the | |
e793c0f7 | 3111 | * socket interface. The value ops->family corresponds to the |
55737fda | 3112 | * socket system call protocol family. |
1da177e4 | 3113 | */ |
f0fd27d4 | 3114 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
3115 | { |
3116 | int err; | |
3117 | ||
3118 | if (ops->family >= NPROTO) { | |
3410f22e | 3119 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
3120 | return -ENOBUFS; |
3121 | } | |
55737fda SH |
3122 | |
3123 | spin_lock(&net_family_lock); | |
190683a9 ED |
3124 | if (rcu_dereference_protected(net_families[ops->family], |
3125 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
3126 | err = -EEXIST; |
3127 | else { | |
cf778b00 | 3128 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
3129 | err = 0; |
3130 | } | |
55737fda SH |
3131 | spin_unlock(&net_family_lock); |
3132 | ||
fe0bdbde | 3133 | pr_info("NET: Registered %s protocol family\n", pf_family_names[ops->family]); |
1da177e4 LT |
3134 | return err; |
3135 | } | |
c6d409cf | 3136 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 3137 | |
55737fda SH |
3138 | /** |
3139 | * sock_unregister - remove a protocol handler | |
3140 | * @family: protocol family to remove | |
3141 | * | |
1da177e4 LT |
3142 | * This function is called by a protocol handler that wants to |
3143 | * remove its address family, and have it unlinked from the | |
55737fda SH |
3144 | * new socket creation. |
3145 | * | |
3146 | * If protocol handler is a module, then it can use module reference | |
3147 | * counts to protect against new references. If protocol handler is not | |
3148 | * a module then it needs to provide its own protection in | |
3149 | * the ops->create routine. | |
1da177e4 | 3150 | */ |
f0fd27d4 | 3151 | void sock_unregister(int family) |
1da177e4 | 3152 | { |
f0fd27d4 | 3153 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 3154 | |
55737fda | 3155 | spin_lock(&net_family_lock); |
a9b3cd7f | 3156 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
3157 | spin_unlock(&net_family_lock); |
3158 | ||
3159 | synchronize_rcu(); | |
3160 | ||
fe0bdbde | 3161 | pr_info("NET: Unregistered %s protocol family\n", pf_family_names[family]); |
1da177e4 | 3162 | } |
c6d409cf | 3163 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 3164 | |
bf2ae2e4 XL |
3165 | bool sock_is_registered(int family) |
3166 | { | |
66b51b0a | 3167 | return family < NPROTO && rcu_access_pointer(net_families[family]); |
bf2ae2e4 XL |
3168 | } |
3169 | ||
77d76ea3 | 3170 | static int __init sock_init(void) |
1da177e4 | 3171 | { |
b3e19d92 | 3172 | int err; |
2ca794e5 EB |
3173 | /* |
3174 | * Initialize the network sysctl infrastructure. | |
3175 | */ | |
3176 | err = net_sysctl_init(); | |
3177 | if (err) | |
3178 | goto out; | |
b3e19d92 | 3179 | |
1da177e4 | 3180 | /* |
89bddce5 | 3181 | * Initialize skbuff SLAB cache |
1da177e4 LT |
3182 | */ |
3183 | skb_init(); | |
1da177e4 LT |
3184 | |
3185 | /* | |
89bddce5 | 3186 | * Initialize the protocols module. |
1da177e4 LT |
3187 | */ |
3188 | ||
3189 | init_inodecache(); | |
b3e19d92 NP |
3190 | |
3191 | err = register_filesystem(&sock_fs_type); | |
3192 | if (err) | |
47260ba9 | 3193 | goto out; |
1da177e4 | 3194 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
3195 | if (IS_ERR(sock_mnt)) { |
3196 | err = PTR_ERR(sock_mnt); | |
3197 | goto out_mount; | |
3198 | } | |
77d76ea3 AK |
3199 | |
3200 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
3201 | */ |
3202 | ||
3203 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
3204 | err = netfilter_init(); |
3205 | if (err) | |
3206 | goto out; | |
1da177e4 | 3207 | #endif |
cbeb321a | 3208 | |
408eccce | 3209 | ptp_classifier_init(); |
c1f19b51 | 3210 | |
b3e19d92 NP |
3211 | out: |
3212 | return err; | |
3213 | ||
3214 | out_mount: | |
3215 | unregister_filesystem(&sock_fs_type); | |
b3e19d92 | 3216 | goto out; |
1da177e4 LT |
3217 | } |
3218 | ||
77d76ea3 AK |
3219 | core_initcall(sock_init); /* early initcall */ |
3220 | ||
1da177e4 LT |
3221 | #ifdef CONFIG_PROC_FS |
3222 | void socket_seq_show(struct seq_file *seq) | |
3223 | { | |
648845ab TZ |
3224 | seq_printf(seq, "sockets: used %d\n", |
3225 | sock_inuse_get(seq->private)); | |
1da177e4 | 3226 | } |
89bddce5 | 3227 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 3228 | |
29c49648 AB |
3229 | /* Handle the fact that while struct ifreq has the same *layout* on |
3230 | * 32/64 for everything but ifreq::ifru_ifmap and ifreq::ifru_data, | |
3231 | * which are handled elsewhere, it still has different *size* due to | |
3232 | * ifreq::ifru_ifmap (which is 16 bytes on 32 bit, 24 bytes on 64-bit, | |
3233 | * resulting in struct ifreq being 32 and 40 bytes respectively). | |
3234 | * As a result, if the struct happens to be at the end of a page and | |
3235 | * the next page isn't readable/writable, we get a fault. To prevent | |
3236 | * that, copy back and forth to the full size. | |
3237 | */ | |
3238 | int get_user_ifreq(struct ifreq *ifr, void __user **ifrdata, void __user *arg) | |
7a229387 | 3239 | { |
29c49648 AB |
3240 | if (in_compat_syscall()) { |
3241 | struct compat_ifreq *ifr32 = (struct compat_ifreq *)ifr; | |
7a229387 | 3242 | |
29c49648 AB |
3243 | memset(ifr, 0, sizeof(*ifr)); |
3244 | if (copy_from_user(ifr32, arg, sizeof(*ifr32))) | |
3245 | return -EFAULT; | |
7a229387 | 3246 | |
29c49648 AB |
3247 | if (ifrdata) |
3248 | *ifrdata = compat_ptr(ifr32->ifr_data); | |
7a229387 | 3249 | |
29c49648 AB |
3250 | return 0; |
3251 | } | |
7a229387 | 3252 | |
29c49648 | 3253 | if (copy_from_user(ifr, arg, sizeof(*ifr))) |
7a229387 AB |
3254 | return -EFAULT; |
3255 | ||
29c49648 AB |
3256 | if (ifrdata) |
3257 | *ifrdata = ifr->ifr_data; | |
3258 | ||
7a229387 AB |
3259 | return 0; |
3260 | } | |
29c49648 | 3261 | EXPORT_SYMBOL(get_user_ifreq); |
7a229387 | 3262 | |
29c49648 | 3263 | int put_user_ifreq(struct ifreq *ifr, void __user *arg) |
7a229387 | 3264 | { |
29c49648 | 3265 | size_t size = sizeof(*ifr); |
7a229387 | 3266 | |
29c49648 AB |
3267 | if (in_compat_syscall()) |
3268 | size = sizeof(struct compat_ifreq); | |
7a229387 | 3269 | |
29c49648 | 3270 | if (copy_to_user(arg, ifr, size)) |
7a229387 AB |
3271 | return -EFAULT; |
3272 | ||
3a7da39d | 3273 | return 0; |
7a229387 | 3274 | } |
29c49648 | 3275 | EXPORT_SYMBOL(put_user_ifreq); |
7a229387 | 3276 | |
89bbfc95 | 3277 | #ifdef CONFIG_COMPAT |
7a50a240 AB |
3278 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
3279 | { | |
7a50a240 | 3280 | compat_uptr_t uptr32; |
44c02a2c AV |
3281 | struct ifreq ifr; |
3282 | void __user *saved; | |
3283 | int err; | |
7a50a240 | 3284 | |
29c49648 | 3285 | if (get_user_ifreq(&ifr, NULL, uifr32)) |
7a50a240 AB |
3286 | return -EFAULT; |
3287 | ||
3288 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
3289 | return -EFAULT; | |
3290 | ||
44c02a2c AV |
3291 | saved = ifr.ifr_settings.ifs_ifsu.raw_hdlc; |
3292 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = compat_ptr(uptr32); | |
7a229387 | 3293 | |
a554bf96 | 3294 | err = dev_ioctl(net, SIOCWANDEV, &ifr, NULL, NULL); |
44c02a2c AV |
3295 | if (!err) { |
3296 | ifr.ifr_settings.ifs_ifsu.raw_hdlc = saved; | |
29c49648 | 3297 | if (put_user_ifreq(&ifr, uifr32)) |
44c02a2c | 3298 | err = -EFAULT; |
ccbd6a5a | 3299 | } |
44c02a2c | 3300 | return err; |
7a229387 AB |
3301 | } |
3302 | ||
590d4693 BH |
3303 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
3304 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 3305 | struct compat_ifreq __user *u_ifreq32) |
7a229387 | 3306 | { |
44c02a2c | 3307 | struct ifreq ifreq; |
a554bf96 | 3308 | void __user *data; |
7a229387 | 3309 | |
d0efb162 PC |
3310 | if (!is_socket_ioctl_cmd(cmd)) |
3311 | return -ENOTTY; | |
a554bf96 | 3312 | if (get_user_ifreq(&ifreq, &data, u_ifreq32)) |
7a229387 | 3313 | return -EFAULT; |
a554bf96 | 3314 | ifreq.ifr_data = data; |
7a229387 | 3315 | |
a554bf96 | 3316 | return dev_ioctl(net, cmd, &ifreq, data, NULL); |
a2116ed2 AB |
3317 | } |
3318 | ||
6b96018b AB |
3319 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3320 | unsigned int cmd, unsigned long arg) | |
3321 | { | |
3322 | void __user *argp = compat_ptr(arg); | |
3323 | struct sock *sk = sock->sk; | |
3324 | struct net *net = sock_net(sk); | |
7a229387 | 3325 | |
6b96018b | 3326 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
88fc023f | 3327 | return sock_ioctl(file, cmd, (unsigned long)argp); |
6b96018b AB |
3328 | |
3329 | switch (cmd) { | |
7a50a240 AB |
3330 | case SIOCWANDEV: |
3331 | return compat_siocwandev(net, argp); | |
0768e170 AB |
3332 | case SIOCGSTAMP_OLD: |
3333 | case SIOCGSTAMPNS_OLD: | |
c7cbdbf2 AB |
3334 | if (!sock->ops->gettstamp) |
3335 | return -ENOIOCTLCMD; | |
0768e170 | 3336 | return sock->ops->gettstamp(sock, argp, cmd == SIOCGSTAMP_OLD, |
c7cbdbf2 AB |
3337 | !COMPAT_USE_64BIT_TIME); |
3338 | ||
dd98d289 | 3339 | case SIOCETHTOOL: |
590d4693 BH |
3340 | case SIOCBONDSLAVEINFOQUERY: |
3341 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3342 | case SIOCSHWTSTAMP: |
fd468c74 | 3343 | case SIOCGHWTSTAMP: |
590d4693 | 3344 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3345 | |
3346 | case FIOSETOWN: | |
3347 | case SIOCSPGRP: | |
3348 | case FIOGETOWN: | |
3349 | case SIOCGPGRP: | |
3350 | case SIOCBRADDBR: | |
3351 | case SIOCBRDELBR: | |
3352 | case SIOCGIFVLAN: | |
3353 | case SIOCSIFVLAN: | |
c62cce2c | 3354 | case SIOCGSKNS: |
0768e170 AB |
3355 | case SIOCGSTAMP_NEW: |
3356 | case SIOCGSTAMPNS_NEW: | |
876f0bf9 | 3357 | case SIOCGIFCONF: |
fd3a4590 RP |
3358 | case SIOCSIFBR: |
3359 | case SIOCGIFBR: | |
6b96018b AB |
3360 | return sock_ioctl(file, cmd, arg); |
3361 | ||
3362 | case SIOCGIFFLAGS: | |
3363 | case SIOCSIFFLAGS: | |
709566d7 AB |
3364 | case SIOCGIFMAP: |
3365 | case SIOCSIFMAP: | |
6b96018b AB |
3366 | case SIOCGIFMETRIC: |
3367 | case SIOCSIFMETRIC: | |
3368 | case SIOCGIFMTU: | |
3369 | case SIOCSIFMTU: | |
3370 | case SIOCGIFMEM: | |
3371 | case SIOCSIFMEM: | |
3372 | case SIOCGIFHWADDR: | |
3373 | case SIOCSIFHWADDR: | |
3374 | case SIOCADDMULTI: | |
3375 | case SIOCDELMULTI: | |
3376 | case SIOCGIFINDEX: | |
6b96018b AB |
3377 | case SIOCGIFADDR: |
3378 | case SIOCSIFADDR: | |
3379 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3380 | case SIOCDIFADDR: |
6b96018b AB |
3381 | case SIOCGIFBRDADDR: |
3382 | case SIOCSIFBRDADDR: | |
3383 | case SIOCGIFDSTADDR: | |
3384 | case SIOCSIFDSTADDR: | |
3385 | case SIOCGIFNETMASK: | |
3386 | case SIOCSIFNETMASK: | |
3387 | case SIOCSIFPFLAGS: | |
3388 | case SIOCGIFPFLAGS: | |
3389 | case SIOCGIFTXQLEN: | |
3390 | case SIOCSIFTXQLEN: | |
3391 | case SIOCBRADDIF: | |
3392 | case SIOCBRDELIF: | |
c6c9fee3 | 3393 | case SIOCGIFNAME: |
9177efd3 AB |
3394 | case SIOCSIFNAME: |
3395 | case SIOCGMIIPHY: | |
3396 | case SIOCGMIIREG: | |
3397 | case SIOCSMIIREG: | |
f92d4fc9 AV |
3398 | case SIOCBONDENSLAVE: |
3399 | case SIOCBONDRELEASE: | |
3400 | case SIOCBONDSETHWADDR: | |
3401 | case SIOCBONDCHANGEACTIVE: | |
6b96018b AB |
3402 | case SIOCSARP: |
3403 | case SIOCGARP: | |
3404 | case SIOCDARP: | |
c7dc504e | 3405 | case SIOCOUTQ: |
9d7bf41f | 3406 | case SIOCOUTQNSD: |
6b96018b | 3407 | case SIOCATMARK: |
63ff03ab | 3408 | return sock_do_ioctl(net, sock, cmd, arg); |
9177efd3 AB |
3409 | } |
3410 | ||
6b96018b AB |
3411 | return -ENOIOCTLCMD; |
3412 | } | |
7a229387 | 3413 | |
95c96174 | 3414 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3415 | unsigned long arg) |
89bbfc95 SP |
3416 | { |
3417 | struct socket *sock = file->private_data; | |
3418 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3419 | struct sock *sk; |
3420 | struct net *net; | |
3421 | ||
3422 | sk = sock->sk; | |
3423 | net = sock_net(sk); | |
89bbfc95 SP |
3424 | |
3425 | if (sock->ops->compat_ioctl) | |
3426 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3427 | ||
87de87d5 DM |
3428 | if (ret == -ENOIOCTLCMD && |
3429 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3430 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3431 | ||
6b96018b AB |
3432 | if (ret == -ENOIOCTLCMD) |
3433 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3434 | ||
89bbfc95 SP |
3435 | return ret; |
3436 | } | |
3437 | #endif | |
3438 | ||
8a3c245c PT |
3439 | /** |
3440 | * kernel_bind - bind an address to a socket (kernel space) | |
3441 | * @sock: socket | |
3442 | * @addr: address | |
3443 | * @addrlen: length of address | |
3444 | * | |
3445 | * Returns 0 or an error. | |
3446 | */ | |
3447 | ||
ac5a488e SS |
3448 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3449 | { | |
3450 | return sock->ops->bind(sock, addr, addrlen); | |
3451 | } | |
c6d409cf | 3452 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e | 3453 | |
8a3c245c PT |
3454 | /** |
3455 | * kernel_listen - move socket to listening state (kernel space) | |
3456 | * @sock: socket | |
3457 | * @backlog: pending connections queue size | |
3458 | * | |
3459 | * Returns 0 or an error. | |
3460 | */ | |
3461 | ||
ac5a488e SS |
3462 | int kernel_listen(struct socket *sock, int backlog) |
3463 | { | |
3464 | return sock->ops->listen(sock, backlog); | |
3465 | } | |
c6d409cf | 3466 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e | 3467 | |
8a3c245c PT |
3468 | /** |
3469 | * kernel_accept - accept a connection (kernel space) | |
3470 | * @sock: listening socket | |
3471 | * @newsock: new connected socket | |
3472 | * @flags: flags | |
3473 | * | |
3474 | * @flags must be SOCK_CLOEXEC, SOCK_NONBLOCK or 0. | |
3475 | * If it fails, @newsock is guaranteed to be %NULL. | |
3476 | * Returns 0 or an error. | |
3477 | */ | |
3478 | ||
ac5a488e SS |
3479 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) |
3480 | { | |
3481 | struct sock *sk = sock->sk; | |
3482 | int err; | |
3483 | ||
3484 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3485 | newsock); | |
3486 | if (err < 0) | |
3487 | goto done; | |
3488 | ||
cdfbabfb | 3489 | err = sock->ops->accept(sock, *newsock, flags, true); |
ac5a488e SS |
3490 | if (err < 0) { |
3491 | sock_release(*newsock); | |
fa8705b0 | 3492 | *newsock = NULL; |
ac5a488e SS |
3493 | goto done; |
3494 | } | |
3495 | ||
3496 | (*newsock)->ops = sock->ops; | |
1b08534e | 3497 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3498 | |
3499 | done: | |
3500 | return err; | |
3501 | } | |
c6d409cf | 3502 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e | 3503 | |
8a3c245c PT |
3504 | /** |
3505 | * kernel_connect - connect a socket (kernel space) | |
3506 | * @sock: socket | |
3507 | * @addr: address | |
3508 | * @addrlen: address length | |
3509 | * @flags: flags (O_NONBLOCK, ...) | |
3510 | * | |
f1dcffcc | 3511 | * For datagram sockets, @addr is the address to which datagrams are sent |
8a3c245c PT |
3512 | * by default, and the only address from which datagrams are received. |
3513 | * For stream sockets, attempts to connect to @addr. | |
3514 | * Returns 0 or an error code. | |
3515 | */ | |
3516 | ||
ac5a488e | 3517 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, |
4768fbcb | 3518 | int flags) |
ac5a488e SS |
3519 | { |
3520 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3521 | } | |
c6d409cf | 3522 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e | 3523 | |
8a3c245c PT |
3524 | /** |
3525 | * kernel_getsockname - get the address which the socket is bound (kernel space) | |
3526 | * @sock: socket | |
3527 | * @addr: address holder | |
3528 | * | |
3529 | * Fills the @addr pointer with the address which the socket is bound. | |
0fc95dec | 3530 | * Returns the length of the address in bytes or an error code. |
8a3c245c PT |
3531 | */ |
3532 | ||
9b2c45d4 | 3533 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3534 | { |
9b2c45d4 | 3535 | return sock->ops->getname(sock, addr, 0); |
ac5a488e | 3536 | } |
c6d409cf | 3537 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e | 3538 | |
8a3c245c | 3539 | /** |
645f0897 | 3540 | * kernel_getpeername - get the address which the socket is connected (kernel space) |
8a3c245c PT |
3541 | * @sock: socket |
3542 | * @addr: address holder | |
3543 | * | |
3544 | * Fills the @addr pointer with the address which the socket is connected. | |
0fc95dec | 3545 | * Returns the length of the address in bytes or an error code. |
8a3c245c PT |
3546 | */ |
3547 | ||
9b2c45d4 | 3548 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr) |
ac5a488e | 3549 | { |
9b2c45d4 | 3550 | return sock->ops->getname(sock, addr, 1); |
ac5a488e | 3551 | } |
c6d409cf | 3552 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e | 3553 | |
8a3c245c PT |
3554 | /** |
3555 | * kernel_sendpage - send a &page through a socket (kernel space) | |
3556 | * @sock: socket | |
3557 | * @page: page | |
3558 | * @offset: page offset | |
3559 | * @size: total size in bytes | |
3560 | * @flags: flags (MSG_DONTWAIT, ...) | |
3561 | * | |
3562 | * Returns the total amount sent in bytes or an error. | |
3563 | */ | |
3564 | ||
ac5a488e SS |
3565 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, |
3566 | size_t size, int flags) | |
3567 | { | |
7b62d31d CL |
3568 | if (sock->ops->sendpage) { |
3569 | /* Warn in case the improper page to zero-copy send */ | |
3570 | WARN_ONCE(!sendpage_ok(page), "improper page for zero-copy send"); | |
ac5a488e | 3571 | return sock->ops->sendpage(sock, page, offset, size, flags); |
7b62d31d | 3572 | } |
ac5a488e SS |
3573 | return sock_no_sendpage(sock, page, offset, size, flags); |
3574 | } | |
c6d409cf | 3575 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e | 3576 | |
8a3c245c PT |
3577 | /** |
3578 | * kernel_sendpage_locked - send a &page through the locked sock (kernel space) | |
3579 | * @sk: sock | |
3580 | * @page: page | |
3581 | * @offset: page offset | |
3582 | * @size: total size in bytes | |
3583 | * @flags: flags (MSG_DONTWAIT, ...) | |
3584 | * | |
3585 | * Returns the total amount sent in bytes or an error. | |
3586 | * Caller must hold @sk. | |
3587 | */ | |
3588 | ||
306b13eb TH |
3589 | int kernel_sendpage_locked(struct sock *sk, struct page *page, int offset, |
3590 | size_t size, int flags) | |
3591 | { | |
3592 | struct socket *sock = sk->sk_socket; | |
3593 | ||
3594 | if (sock->ops->sendpage_locked) | |
3595 | return sock->ops->sendpage_locked(sk, page, offset, size, | |
3596 | flags); | |
3597 | ||
3598 | return sock_no_sendpage_locked(sk, page, offset, size, flags); | |
3599 | } | |
3600 | EXPORT_SYMBOL(kernel_sendpage_locked); | |
3601 | ||
8a3c245c | 3602 | /** |
645f0897 | 3603 | * kernel_sock_shutdown - shut down part of a full-duplex connection (kernel space) |
8a3c245c PT |
3604 | * @sock: socket |
3605 | * @how: connection part | |
3606 | * | |
3607 | * Returns 0 or an error. | |
3608 | */ | |
3609 | ||
91cf45f0 TM |
3610 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3611 | { | |
3612 | return sock->ops->shutdown(sock, how); | |
3613 | } | |
91cf45f0 | 3614 | EXPORT_SYMBOL(kernel_sock_shutdown); |
113c3075 | 3615 | |
8a3c245c PT |
3616 | /** |
3617 | * kernel_sock_ip_overhead - returns the IP overhead imposed by a socket | |
3618 | * @sk: socket | |
3619 | * | |
3620 | * This routine returns the IP overhead imposed by a socket i.e. | |
3621 | * the length of the underlying IP header, depending on whether | |
3622 | * this is an IPv4 or IPv6 socket and the length from IP options turned | |
3623 | * on at the socket. Assumes that the caller has a lock on the socket. | |
113c3075 | 3624 | */ |
8a3c245c | 3625 | |
113c3075 P |
3626 | u32 kernel_sock_ip_overhead(struct sock *sk) |
3627 | { | |
3628 | struct inet_sock *inet; | |
3629 | struct ip_options_rcu *opt; | |
3630 | u32 overhead = 0; | |
113c3075 P |
3631 | #if IS_ENABLED(CONFIG_IPV6) |
3632 | struct ipv6_pinfo *np; | |
3633 | struct ipv6_txoptions *optv6 = NULL; | |
3634 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3635 | ||
3636 | if (!sk) | |
3637 | return overhead; | |
3638 | ||
113c3075 P |
3639 | switch (sk->sk_family) { |
3640 | case AF_INET: | |
3641 | inet = inet_sk(sk); | |
3642 | overhead += sizeof(struct iphdr); | |
3643 | opt = rcu_dereference_protected(inet->inet_opt, | |
614d79c0 | 3644 | sock_owned_by_user(sk)); |
113c3075 P |
3645 | if (opt) |
3646 | overhead += opt->opt.optlen; | |
3647 | return overhead; | |
3648 | #if IS_ENABLED(CONFIG_IPV6) | |
3649 | case AF_INET6: | |
3650 | np = inet6_sk(sk); | |
3651 | overhead += sizeof(struct ipv6hdr); | |
3652 | if (np) | |
3653 | optv6 = rcu_dereference_protected(np->opt, | |
614d79c0 | 3654 | sock_owned_by_user(sk)); |
113c3075 P |
3655 | if (optv6) |
3656 | overhead += (optv6->opt_flen + optv6->opt_nflen); | |
3657 | return overhead; | |
3658 | #endif /* IS_ENABLED(CONFIG_IPV6) */ | |
3659 | default: /* Returns 0 overhead if the socket is not ipv4 or ipv6 */ | |
3660 | return overhead; | |
3661 | } | |
3662 | } | |
3663 | EXPORT_SYMBOL(kernel_sock_ip_overhead); |