blob: fe20c319a0bb37dd475b7a8a32dccd57ac4e15d9 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * NET An implementation of the SOCKET network access protocol.
3 *
4 * Version: @(#)socket.c 1.1.93 18/02/95
5 *
6 * Authors: Orest Zborowski, <obz@Kodak.COM>
Jesper Juhl02c30a82005-05-05 16:16:16 -07007 * Ross Biro
Linus Torvalds1da177e2005-04-16 15:20:36 -07008 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
9 *
10 * Fixes:
11 * Anonymous : NOTSOCK/BADF cleanup. Error fix in
12 * shutdown()
13 * Alan Cox : verify_area() fixes
14 * Alan Cox : Removed DDI
15 * Jonathan Kamens : SOCK_DGRAM reconnect bug
16 * Alan Cox : Moved a load of checks to the very
17 * top level.
18 * Alan Cox : Move address structures to/from user
19 * mode above the protocol layers.
20 * Rob Janssen : Allow 0 length sends.
21 * Alan Cox : Asynchronous I/O support (cribbed from the
22 * tty drivers).
23 * Niibe Yutaka : Asynchronous I/O for writes (4.4BSD style)
24 * Jeff Uphoff : Made max number of sockets command-line
25 * configurable.
26 * Matti Aarnio : Made the number of sockets dynamic,
27 * to be allocated when needed, and mr.
28 * Uphoff's max is used as max to be
29 * allowed to allocate.
30 * Linus : Argh. removed all the socket allocation
31 * altogether: it's in the inode now.
32 * Alan Cox : Made sock_alloc()/sock_release() public
33 * for NetROM and future kernel nfsd type
34 * stuff.
35 * Alan Cox : sendmsg/recvmsg basics.
36 * Tom Dyas : Export net symbols.
37 * Marcin Dalecki : Fixed problems with CONFIG_NET="n".
38 * Alan Cox : Added thread locking to sys_* calls
39 * for sockets. May have errors at the
40 * moment.
41 * Kevin Buhr : Fixed the dumb errors in the above.
42 * Andi Kleen : Some small cleanups, optimizations,
43 * and fixed a copy_from_user() bug.
44 * Tigran Aivazian : sys_send(args) calls sys_sendto(args, NULL, 0)
Stephen Hemminger89bddce2006-09-01 00:19:31 -070045 * Tigran Aivazian : Made listen(2) backlog sanity checks
Linus Torvalds1da177e2005-04-16 15:20:36 -070046 * protocol-independent
47 *
48 *
49 * This program is free software; you can redistribute it and/or
50 * modify it under the terms of the GNU General Public License
51 * as published by the Free Software Foundation; either version
52 * 2 of the License, or (at your option) any later version.
53 *
54 *
55 * This module is effectively the top level interface to the BSD socket
Stephen Hemminger89bddce2006-09-01 00:19:31 -070056 * paradigm.
Linus Torvalds1da177e2005-04-16 15:20:36 -070057 *
58 * Based upon Swansea University Computer Society NET3.039
59 */
60
Linus Torvalds1da177e2005-04-16 15:20:36 -070061#include <linux/mm.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070062#include <linux/socket.h>
63#include <linux/file.h>
64#include <linux/net.h>
65#include <linux/interrupt.h>
Ulrich Drepperaaca0bd2008-07-23 21:29:20 -070066#include <linux/thread_info.h>
Stephen Hemminger55737fd2006-09-01 00:23:39 -070067#include <linux/rcupdate.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070068#include <linux/netdevice.h>
69#include <linux/proc_fs.h>
70#include <linux/seq_file.h>
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -080071#include <linux/mutex.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070072#include <linux/if_bridge.h>
Arnaldo Carvalho de Melo20380732005-08-16 02:18:02 -030073#include <linux/if_frad.h>
74#include <linux/if_vlan.h>
Daniel Borkmann408eccc2014-04-01 16:20:23 +020075#include <linux/ptp_classify.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070076#include <linux/init.h>
77#include <linux/poll.h>
78#include <linux/cache.h>
79#include <linux/module.h>
80#include <linux/highmem.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070081#include <linux/mount.h>
82#include <linux/security.h>
83#include <linux/syscalls.h>
84#include <linux/compat.h>
85#include <linux/kmod.h>
David Woodhouse3ec3b2f2005-05-17 12:08:48 +010086#include <linux/audit.h>
Adrian Bunkd86b5e02006-01-21 00:46:55 +010087#include <linux/wireless.h>
Eric W. Biederman1b8d7ae2007-10-08 23:24:22 -070088#include <linux/nsproxy.h>
Nick Black1fd7317d2009-09-22 16:43:33 -070089#include <linux/magic.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +090090#include <linux/slab.h>
Masatake YAMATO600e1772012-08-29 10:44:29 +000091#include <linux/xattr.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070092
93#include <asm/uaccess.h>
94#include <asm/unistd.h>
95
96#include <net/compat.h>
David S. Miller87de87d2008-06-03 09:14:03 -070097#include <net/wext.h>
Herbert Xuf8451722010-05-24 00:12:34 -070098#include <net/cls_cgroup.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070099
100#include <net/sock.h>
101#include <linux/netfilter.h>
102
Arnd Bergmann6b960182009-11-06 23:10:54 -0800103#include <linux/if_tun.h>
104#include <linux/ipv6_route.h>
105#include <linux/route.h>
Arnd Bergmann6b960182009-11-06 23:10:54 -0800106#include <linux/sockios.h>
107#include <linux/atalk.h>
Eliezer Tamir076bb0c2013-07-10 17:13:17 +0300108#include <net/busy_poll.h>
Willem de Bruijnf24b9be2014-08-04 22:11:45 -0400109#include <linux/errqueue.h>
Eliezer Tamir06021292013-06-10 11:39:50 +0300110
Cong Wange0d10952013-08-01 11:10:25 +0800111#ifdef CONFIG_NET_RX_BUSY_POLL
Eliezer Tamir64b0dc52013-07-10 17:13:36 +0300112unsigned int sysctl_net_busy_read __read_mostly;
113unsigned int sysctl_net_busy_poll __read_mostly;
Eliezer Tamir06021292013-06-10 11:39:50 +0300114#endif
Arnd Bergmann6b960182009-11-06 23:10:54 -0800115
Linus Torvalds1da177e2005-04-16 15:20:36 -0700116static int sock_no_open(struct inode *irrelevant, struct file *dontcare);
Badari Pulavarty027445c2006-09-30 23:28:46 -0700117static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov,
118 unsigned long nr_segs, loff_t pos);
119static ssize_t sock_aio_write(struct kiocb *iocb, const struct iovec *iov,
120 unsigned long nr_segs, loff_t pos);
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700121static int sock_mmap(struct file *file, struct vm_area_struct *vma);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700122
123static int sock_close(struct inode *inode, struct file *file);
124static unsigned int sock_poll(struct file *file,
125 struct poll_table_struct *wait);
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700126static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
Shaun Pereira89bbfc92006-03-21 23:58:08 -0800127#ifdef CONFIG_COMPAT
128static long compat_sock_ioctl(struct file *file,
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700129 unsigned int cmd, unsigned long arg);
Shaun Pereira89bbfc92006-03-21 23:58:08 -0800130#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700131static int sock_fasync(int fd, struct file *filp, int on);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700132static ssize_t sock_sendpage(struct file *file, struct page *page,
133 int offset, size_t size, loff_t *ppos, int more);
Jens Axboe9c55e012007-11-06 23:30:13 -0800134static ssize_t sock_splice_read(struct file *file, loff_t *ppos,
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700135 struct pipe_inode_info *pipe, size_t len,
Jens Axboe9c55e012007-11-06 23:30:13 -0800136 unsigned int flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700137
Linus Torvalds1da177e2005-04-16 15:20:36 -0700138/*
139 * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear
140 * in the operation structures but are done directly via the socketcall() multiplexor.
141 */
142
Arjan van de Venda7071d2007-02-12 00:55:36 -0800143static const struct file_operations socket_file_ops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700144 .owner = THIS_MODULE,
145 .llseek = no_llseek,
146 .aio_read = sock_aio_read,
147 .aio_write = sock_aio_write,
148 .poll = sock_poll,
149 .unlocked_ioctl = sock_ioctl,
Shaun Pereira89bbfc92006-03-21 23:58:08 -0800150#ifdef CONFIG_COMPAT
151 .compat_ioctl = compat_sock_ioctl,
152#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700153 .mmap = sock_mmap,
154 .open = sock_no_open, /* special open code to disallow open via /proc */
155 .release = sock_close,
156 .fasync = sock_fasync,
Jens Axboe5274f052006-03-30 15:15:30 +0200157 .sendpage = sock_sendpage,
158 .splice_write = generic_splice_sendpage,
Jens Axboe9c55e012007-11-06 23:30:13 -0800159 .splice_read = sock_splice_read,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700160};
161
162/*
163 * The protocol list. Each protocol is registered in here.
164 */
165
Linus Torvalds1da177e2005-04-16 15:20:36 -0700166static DEFINE_SPINLOCK(net_family_lock);
Eric Dumazet190683a2010-11-10 10:50:44 +0000167static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700168
Linus Torvalds1da177e2005-04-16 15:20:36 -0700169/*
170 * Statistics counters of the socket lists
171 */
172
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700173static DEFINE_PER_CPU(int, sockets_in_use);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700174
175/*
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700176 * Support routines.
177 * Move socket addresses back and forth across the kernel/user
178 * divide and look after the messy bits.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700179 */
180
Linus Torvalds1da177e2005-04-16 15:20:36 -0700181/**
182 * move_addr_to_kernel - copy a socket address into kernel space
183 * @uaddr: Address in user space
184 * @kaddr: Address in kernel space
185 * @ulen: Length in user space
186 *
187 * The address is copied into kernel space. If the provided address is
188 * too long an error code of -EINVAL is returned. If the copy gives
189 * invalid addresses -EFAULT is returned. On a success 0 is returned.
190 */
191
Maciej Żenczykowski43db3622012-03-11 12:51:50 +0000192int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700193{
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -0700194 if (ulen < 0 || ulen > sizeof(struct sockaddr_storage))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700195 return -EINVAL;
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700196 if (ulen == 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700197 return 0;
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700198 if (copy_from_user(kaddr, uaddr, ulen))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700199 return -EFAULT;
David Woodhouse3ec3b2f2005-05-17 12:08:48 +0100200 return audit_sockaddr(ulen, kaddr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700201}
202
203/**
204 * move_addr_to_user - copy an address to user space
205 * @kaddr: kernel space address
206 * @klen: length of address in kernel
207 * @uaddr: user space address
208 * @ulen: pointer to user length field
209 *
210 * The value pointed to by ulen on entry is the buffer length available.
211 * This is overwritten with the buffer space used. -EINVAL is returned
212 * if an overlong buffer is specified or a negative buffer size. -EFAULT
213 * is returned if either the buffer or the length field are not
214 * accessible.
215 * After copying the data up to the limit the user specifies, the true
216 * length of the data is written over the length limit the user
217 * specified. Zero is returned for a success.
218 */
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700219
Maciej Żenczykowski43db3622012-03-11 12:51:50 +0000220static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen,
stephen hemminger11165f12010-10-18 14:27:29 +0000221 void __user *uaddr, int __user *ulen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700222{
223 int err;
224 int len;
225
Hannes Frederic Sowa68c6beb2013-11-21 03:14:34 +0100226 BUG_ON(klen > sizeof(struct sockaddr_storage));
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700227 err = get_user(len, ulen);
228 if (err)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700229 return err;
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700230 if (len > klen)
231 len = klen;
Hannes Frederic Sowa68c6beb2013-11-21 03:14:34 +0100232 if (len < 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700233 return -EINVAL;
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700234 if (len) {
Steve Grubbd6fe3942006-03-30 12:20:22 -0500235 if (audit_sockaddr(klen, kaddr))
236 return -ENOMEM;
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700237 if (copy_to_user(uaddr, kaddr, len))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700238 return -EFAULT;
239 }
240 /*
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700241 * "fromlen shall refer to the value before truncation.."
242 * 1003.1g
Linus Torvalds1da177e2005-04-16 15:20:36 -0700243 */
244 return __put_user(klen, ulen);
245}
246
Christoph Lametere18b8902006-12-06 20:33:20 -0800247static struct kmem_cache *sock_inode_cachep __read_mostly;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700248
249static struct inode *sock_alloc_inode(struct super_block *sb)
250{
251 struct socket_alloc *ei;
Eric Dumazeteaefd1102011-02-18 03:26:36 +0000252 struct socket_wq *wq;
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700253
Christoph Lametere94b1762006-12-06 20:33:17 -0800254 ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700255 if (!ei)
256 return NULL;
Eric Dumazeteaefd1102011-02-18 03:26:36 +0000257 wq = kmalloc(sizeof(*wq), GFP_KERNEL);
258 if (!wq) {
Eric Dumazet43815482010-04-29 11:01:49 +0000259 kmem_cache_free(sock_inode_cachep, ei);
260 return NULL;
261 }
Eric Dumazeteaefd1102011-02-18 03:26:36 +0000262 init_waitqueue_head(&wq->wait);
263 wq->fasync_list = NULL;
264 RCU_INIT_POINTER(ei->socket.wq, wq);
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700265
Linus Torvalds1da177e2005-04-16 15:20:36 -0700266 ei->socket.state = SS_UNCONNECTED;
267 ei->socket.flags = 0;
268 ei->socket.ops = NULL;
269 ei->socket.sk = NULL;
270 ei->socket.file = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700271
272 return &ei->vfs_inode;
273}
274
275static void sock_destroy_inode(struct inode *inode)
276{
Eric Dumazet43815482010-04-29 11:01:49 +0000277 struct socket_alloc *ei;
Eric Dumazeteaefd1102011-02-18 03:26:36 +0000278 struct socket_wq *wq;
Eric Dumazet43815482010-04-29 11:01:49 +0000279
280 ei = container_of(inode, struct socket_alloc, vfs_inode);
Eric Dumazeteaefd1102011-02-18 03:26:36 +0000281 wq = rcu_dereference_protected(ei->socket.wq, 1);
Lai Jiangshan61845222011-03-18 12:10:25 +0800282 kfree_rcu(wq, rcu);
Eric Dumazet43815482010-04-29 11:01:49 +0000283 kmem_cache_free(sock_inode_cachep, ei);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700284}
285
Alexey Dobriyan51cc5062008-07-25 19:45:34 -0700286static void init_once(void *foo)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700287{
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700288 struct socket_alloc *ei = (struct socket_alloc *)foo;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700289
Christoph Lametera35afb82007-05-16 22:10:57 -0700290 inode_init_once(&ei->vfs_inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700291}
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700292
Linus Torvalds1da177e2005-04-16 15:20:36 -0700293static int init_inodecache(void)
294{
295 sock_inode_cachep = kmem_cache_create("sock_inode_cache",
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700296 sizeof(struct socket_alloc),
297 0,
298 (SLAB_HWCACHE_ALIGN |
299 SLAB_RECLAIM_ACCOUNT |
300 SLAB_MEM_SPREAD),
Paul Mundt20c2df82007-07-20 10:11:58 +0900301 init_once);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700302 if (sock_inode_cachep == NULL)
303 return -ENOMEM;
304 return 0;
305}
306
Alexey Dobriyanb87221d2009-09-21 17:01:09 -0700307static const struct super_operations sockfs_ops = {
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700308 .alloc_inode = sock_alloc_inode,
309 .destroy_inode = sock_destroy_inode,
310 .statfs = simple_statfs,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700311};
312
Eric Dumazetc23fbb62007-05-08 00:26:18 -0700313/*
314 * sockfs_dname() is called from d_path().
315 */
316static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen)
317{
318 return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]",
319 dentry->d_inode->i_ino);
320}
321
Al Viro3ba13d12009-02-20 06:02:22 +0000322static const struct dentry_operations sockfs_dentry_operations = {
Eric Dumazetc23fbb62007-05-08 00:26:18 -0700323 .d_dname = sockfs_dname,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700324};
325
Al Viroc74a1cb2011-01-12 16:59:34 -0500326static struct dentry *sockfs_mount(struct file_system_type *fs_type,
327 int flags, const char *dev_name, void *data)
328{
329 return mount_pseudo(fs_type, "socket:", &sockfs_ops,
330 &sockfs_dentry_operations, SOCKFS_MAGIC);
331}
332
333static struct vfsmount *sock_mnt __read_mostly;
334
335static struct file_system_type sock_fs_type = {
336 .name = "sockfs",
337 .mount = sockfs_mount,
338 .kill_sb = kill_anon_super,
339};
340
Linus Torvalds1da177e2005-04-16 15:20:36 -0700341/*
342 * Obtains the first available file descriptor and sets it up for use.
343 *
David S. Miller39d8c1b2006-03-20 17:13:49 -0800344 * These functions create file structures and maps them to fd space
345 * of the current process. On success it returns file descriptor
Linus Torvalds1da177e2005-04-16 15:20:36 -0700346 * and file struct implicitly stored in sock->file.
347 * Note that another thread may close file descriptor before we return
348 * from this function. We use the fact that now we do not refer
349 * to socket after mapping. If one day we will need it, this
350 * function will increment ref. count on file by 1.
351 *
352 * In any case returned fd MAY BE not valid!
353 * This race condition is unavoidable
354 * with shared fd spaces, we cannot solve it inside kernel,
355 * but we take care of internal coherence yet.
356 */
357
Linus Torvaldsaab174f2012-10-02 20:25:04 -0700358struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700359{
Al Viro7cbe66b2009-08-05 19:59:08 +0400360 struct qstr name = { .name = "" };
Al Viro2c48b9c2009-08-09 00:52:35 +0400361 struct path path;
Al Viro7cbe66b2009-08-05 19:59:08 +0400362 struct file *file;
David S. Miller39d8c1b2006-03-20 17:13:49 -0800363
Masatake YAMATO600e1772012-08-29 10:44:29 +0000364 if (dname) {
365 name.name = dname;
366 name.len = strlen(name.name);
367 } else if (sock->sk) {
368 name.name = sock->sk->sk_prot_creator->name;
369 name.len = strlen(name.name);
370 }
Nick Piggin4b936882011-01-07 17:50:07 +1100371 path.dentry = d_alloc_pseudo(sock_mnt->mnt_sb, &name);
Al Viro28407632012-08-17 23:54:15 -0400372 if (unlikely(!path.dentry))
373 return ERR_PTR(-ENOMEM);
Al Viro2c48b9c2009-08-09 00:52:35 +0400374 path.mnt = mntget(sock_mnt);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700375
Al Viro2c48b9c2009-08-09 00:52:35 +0400376 d_instantiate(path.dentry, SOCK_INODE(sock));
Al Virocc3808f2009-08-06 09:43:59 +0400377 SOCK_INODE(sock)->i_fop = &socket_file_ops;
378
Al Viro2c48b9c2009-08-09 00:52:35 +0400379 file = alloc_file(&path, FMODE_READ | FMODE_WRITE,
Al Virocc3808f2009-08-06 09:43:59 +0400380 &socket_file_ops);
Anatol Pomozov39b652522012-09-12 20:11:55 -0700381 if (unlikely(IS_ERR(file))) {
Al Virocc3808f2009-08-06 09:43:59 +0400382 /* drop dentry, keep inode */
Al Viro7de9c6ee2010-10-23 11:11:40 -0400383 ihold(path.dentry->d_inode);
Al Viro2c48b9c2009-08-09 00:52:35 +0400384 path_put(&path);
Anatol Pomozov39b652522012-09-12 20:11:55 -0700385 return file;
Al Virocc3808f2009-08-06 09:43:59 +0400386 }
David S. Miller39d8c1b2006-03-20 17:13:49 -0800387
388 sock->file = file;
Ulrich Drepper77d27202008-07-23 21:29:35 -0700389 file->f_flags = O_RDWR | (flags & O_NONBLOCK);
David S. Miller39d8c1b2006-03-20 17:13:49 -0800390 file->private_data = sock;
Al Viro28407632012-08-17 23:54:15 -0400391 return file;
David S. Miller39d8c1b2006-03-20 17:13:49 -0800392}
Al Viro56b31d12012-08-18 00:25:51 -0400393EXPORT_SYMBOL(sock_alloc_file);
David S. Miller39d8c1b2006-03-20 17:13:49 -0800394
Al Viro56b31d12012-08-18 00:25:51 -0400395static int sock_map_fd(struct socket *sock, int flags)
David S. Miller39d8c1b2006-03-20 17:13:49 -0800396{
397 struct file *newfile;
Al Viro28407632012-08-17 23:54:15 -0400398 int fd = get_unused_fd_flags(flags);
399 if (unlikely(fd < 0))
400 return fd;
David S. Miller39d8c1b2006-03-20 17:13:49 -0800401
Linus Torvaldsaab174f2012-10-02 20:25:04 -0700402 newfile = sock_alloc_file(sock, flags, NULL);
Al Viro28407632012-08-17 23:54:15 -0400403 if (likely(!IS_ERR(newfile))) {
David S. Miller39d8c1b2006-03-20 17:13:49 -0800404 fd_install(fd, newfile);
Al Viro28407632012-08-17 23:54:15 -0400405 return fd;
406 }
Al Viro7cbe66b2009-08-05 19:59:08 +0400407
Al Viro28407632012-08-17 23:54:15 -0400408 put_unused_fd(fd);
409 return PTR_ERR(newfile);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700410}
411
John Fastabend406a3c62012-07-20 10:39:25 +0000412struct socket *sock_from_file(struct file *file, int *err)
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -0800413{
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -0800414 if (file->f_op == &socket_file_ops)
415 return file->private_data; /* set in sock_map_fd */
416
Eric Dumazet23bb80d2007-02-08 14:59:57 -0800417 *err = -ENOTSOCK;
418 return NULL;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -0800419}
John Fastabend406a3c62012-07-20 10:39:25 +0000420EXPORT_SYMBOL(sock_from_file);
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -0800421
Linus Torvalds1da177e2005-04-16 15:20:36 -0700422/**
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700423 * sockfd_lookup - Go from a file number to its socket slot
Linus Torvalds1da177e2005-04-16 15:20:36 -0700424 * @fd: file handle
425 * @err: pointer to an error code return
426 *
427 * The file handle passed in is locked and the socket it is bound
428 * too is returned. If an error occurs the err pointer is overwritten
429 * with a negative errno code and NULL is returned. The function checks
430 * for both invalid handles and passing a handle which is not a socket.
431 *
432 * On a success the socket object pointer is returned.
433 */
434
435struct socket *sockfd_lookup(int fd, int *err)
436{
437 struct file *file;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700438 struct socket *sock;
439
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700440 file = fget(fd);
441 if (!file) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700442 *err = -EBADF;
443 return NULL;
444 }
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700445
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -0800446 sock = sock_from_file(file, err);
447 if (!sock)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700448 fput(file);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700449 return sock;
450}
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700451EXPORT_SYMBOL(sockfd_lookup);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700452
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -0800453static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed)
454{
Al Viro00e188e2014-03-03 23:48:18 -0500455 struct fd f = fdget(fd);
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -0800456 struct socket *sock;
457
Hua Zhong36725582006-04-19 15:25:02 -0700458 *err = -EBADF;
Al Viro00e188e2014-03-03 23:48:18 -0500459 if (f.file) {
460 sock = sock_from_file(f.file, err);
461 if (likely(sock)) {
462 *fput_needed = f.flags;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -0800463 return sock;
Al Viro00e188e2014-03-03 23:48:18 -0500464 }
465 fdput(f);
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -0800466 }
467 return NULL;
468}
469
Masatake YAMATO600e1772012-08-29 10:44:29 +0000470#define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname"
471#define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX)
472#define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1)
473static ssize_t sockfs_getxattr(struct dentry *dentry,
474 const char *name, void *value, size_t size)
475{
476 const char *proto_name;
477 size_t proto_size;
478 int error;
479
480 error = -ENODATA;
481 if (!strncmp(name, XATTR_NAME_SOCKPROTONAME, XATTR_NAME_SOCKPROTONAME_LEN)) {
482 proto_name = dentry->d_name.name;
483 proto_size = strlen(proto_name);
484
485 if (value) {
486 error = -ERANGE;
487 if (proto_size + 1 > size)
488 goto out;
489
490 strncpy(value, proto_name, proto_size + 1);
491 }
492 error = proto_size + 1;
493 }
494
495out:
496 return error;
497}
498
499static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer,
500 size_t size)
501{
502 ssize_t len;
503 ssize_t used = 0;
504
505 len = security_inode_listsecurity(dentry->d_inode, buffer, size);
506 if (len < 0)
507 return len;
508 used += len;
509 if (buffer) {
510 if (size < used)
511 return -ERANGE;
512 buffer += len;
513 }
514
515 len = (XATTR_NAME_SOCKPROTONAME_LEN + 1);
516 used += len;
517 if (buffer) {
518 if (size < used)
519 return -ERANGE;
520 memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len);
521 buffer += len;
522 }
523
524 return used;
525}
526
527static const struct inode_operations sockfs_inode_ops = {
528 .getxattr = sockfs_getxattr,
529 .listxattr = sockfs_listxattr,
530};
531
Linus Torvalds1da177e2005-04-16 15:20:36 -0700532/**
533 * sock_alloc - allocate a socket
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700534 *
Linus Torvalds1da177e2005-04-16 15:20:36 -0700535 * Allocate a new inode and socket object. The two are bound together
536 * and initialised. The socket is then returned. If we are out of inodes
537 * NULL is returned.
538 */
539
540static struct socket *sock_alloc(void)
541{
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700542 struct inode *inode;
543 struct socket *sock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700544
Eric Dumazeta209dfc2011-07-26 11:36:34 +0200545 inode = new_inode_pseudo(sock_mnt->mnt_sb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700546 if (!inode)
547 return NULL;
548
549 sock = SOCKET_I(inode);
550
Eric Dumazet29a020d2009-09-15 02:39:20 -0700551 kmemcheck_annotate_bitfield(sock, type);
Christoph Hellwig85fe4022010-10-23 11:19:54 -0400552 inode->i_ino = get_next_ino();
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700553 inode->i_mode = S_IFSOCK | S_IRWXUGO;
David Howells8192b0c2008-11-14 10:39:10 +1100554 inode->i_uid = current_fsuid();
555 inode->i_gid = current_fsgid();
Masatake YAMATO600e1772012-08-29 10:44:29 +0000556 inode->i_op = &sockfs_inode_ops;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700557
Alex Shi19e8d692012-05-14 14:15:31 -0700558 this_cpu_add(sockets_in_use, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700559 return sock;
560}
561
562/*
563 * In theory you can't get an open on this inode, but /proc provides
564 * a back door. Remember to keep it shut otherwise you'll let the
565 * creepy crawlies in.
566 */
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700567
Linus Torvalds1da177e2005-04-16 15:20:36 -0700568static int sock_no_open(struct inode *irrelevant, struct file *dontcare)
569{
570 return -ENXIO;
571}
572
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -0800573const struct file_operations bad_sock_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700574 .owner = THIS_MODULE,
575 .open = sock_no_open,
Arnd Bergmann6038f372010-08-15 18:52:59 +0200576 .llseek = noop_llseek,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700577};
578
579/**
580 * sock_release - close a socket
581 * @sock: socket to close
582 *
583 * The socket is released from the protocol stack if it has a release
584 * callback, and the inode is then released if the socket is bound to
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700585 * an inode not a file.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700586 */
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700587
Linus Torvalds1da177e2005-04-16 15:20:36 -0700588void sock_release(struct socket *sock)
589{
590 if (sock->ops) {
591 struct module *owner = sock->ops->owner;
592
593 sock->ops->release(sock);
594 sock->ops = NULL;
595 module_put(owner);
596 }
597
Eric Dumazeteaefd1102011-02-18 03:26:36 +0000598 if (rcu_dereference_protected(sock->wq, 1)->fasync_list)
Yang Yingliang3410f222014-02-12 17:09:55 +0800599 pr_err("%s: fasync list not empty!\n", __func__);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700600
Mikulas Patockab09e7862012-07-19 06:13:36 +0000601 if (test_bit(SOCK_EXTERNALLY_ALLOCATED, &sock->flags))
602 return;
603
Alex Shi19e8d692012-05-14 14:15:31 -0700604 this_cpu_sub(sockets_in_use, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700605 if (!sock->file) {
606 iput(SOCK_INODE(sock));
607 return;
608 }
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700609 sock->file = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700610}
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700611EXPORT_SYMBOL(sock_release);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700612
Willem de Bruijn67cc0d42014-09-08 19:58:58 -0400613void __sock_tx_timestamp(const struct sock *sk, __u8 *tx_flags)
Patrick Ohly20d49472009-02-12 05:03:38 +0000614{
Eric Dumazet140c55d2014-08-06 11:49:29 +0200615 u8 flags = *tx_flags;
616
Willem de Bruijnb9f40e22014-08-04 22:11:46 -0400617 if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_HARDWARE)
Eric Dumazet140c55d2014-08-06 11:49:29 +0200618 flags |= SKBTX_HW_TSTAMP;
619
Willem de Bruijnb9f40e22014-08-04 22:11:46 -0400620 if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_SOFTWARE)
Eric Dumazet140c55d2014-08-06 11:49:29 +0200621 flags |= SKBTX_SW_TSTAMP;
622
Willem de Bruijne7fd2882014-08-04 22:11:48 -0400623 if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_SCHED)
Eric Dumazet140c55d2014-08-06 11:49:29 +0200624 flags |= SKBTX_SCHED_TSTAMP;
625
Willem de Bruijne1c8a602014-08-04 22:11:50 -0400626 if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_ACK)
Eric Dumazet140c55d2014-08-06 11:49:29 +0200627 flags |= SKBTX_ACK_TSTAMP;
Willem de Bruijne7fd2882014-08-04 22:11:48 -0400628
Eric Dumazet140c55d2014-08-06 11:49:29 +0200629 *tx_flags = flags;
Patrick Ohly20d49472009-02-12 05:03:38 +0000630}
Willem de Bruijn67cc0d42014-09-08 19:58:58 -0400631EXPORT_SYMBOL(__sock_tx_timestamp);
Patrick Ohly20d49472009-02-12 05:03:38 +0000632
Anton Blanchard228e5482011-05-02 20:21:35 +0000633static inline int __sock_sendmsg_nosec(struct kiocb *iocb, struct socket *sock,
634 struct msghdr *msg, size_t size)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700635{
636 struct sock_iocb *si = kiocb_to_siocb(iocb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700637
638 si->sock = sock;
639 si->scm = NULL;
640 si->msg = msg;
641 si->size = size;
642
Linus Torvalds1da177e2005-04-16 15:20:36 -0700643 return sock->ops->sendmsg(iocb, sock, msg, size);
644}
645
Anton Blanchard228e5482011-05-02 20:21:35 +0000646static inline int __sock_sendmsg(struct kiocb *iocb, struct socket *sock,
647 struct msghdr *msg, size_t size)
648{
649 int err = security_socket_sendmsg(sock, msg, size);
650
651 return err ?: __sock_sendmsg_nosec(iocb, sock, msg, size);
652}
653
Linus Torvalds1da177e2005-04-16 15:20:36 -0700654int sock_sendmsg(struct socket *sock, struct msghdr *msg, size_t size)
655{
656 struct kiocb iocb;
657 struct sock_iocb siocb;
658 int ret;
659
660 init_sync_kiocb(&iocb, NULL);
661 iocb.private = &siocb;
662 ret = __sock_sendmsg(&iocb, sock, msg, size);
663 if (-EIOCBQUEUED == ret)
664 ret = wait_on_sync_kiocb(&iocb);
665 return ret;
666}
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700667EXPORT_SYMBOL(sock_sendmsg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700668
Eric Dumazet894dc242011-07-26 02:39:41 +0000669static int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg, size_t size)
Anton Blanchard228e5482011-05-02 20:21:35 +0000670{
671 struct kiocb iocb;
672 struct sock_iocb siocb;
673 int ret;
674
675 init_sync_kiocb(&iocb, NULL);
676 iocb.private = &siocb;
677 ret = __sock_sendmsg_nosec(&iocb, sock, msg, size);
678 if (-EIOCBQUEUED == ret)
679 ret = wait_on_sync_kiocb(&iocb);
680 return ret;
681}
682
Linus Torvalds1da177e2005-04-16 15:20:36 -0700683int kernel_sendmsg(struct socket *sock, struct msghdr *msg,
684 struct kvec *vec, size_t num, size_t size)
685{
686 mm_segment_t oldfs = get_fs();
687 int result;
688
689 set_fs(KERNEL_DS);
690 /*
691 * the following is safe, since for compiler definitions of kvec and
692 * iovec are identical, yielding the same in-core layout and alignment
693 */
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700694 msg->msg_iov = (struct iovec *)vec;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700695 msg->msg_iovlen = num;
696 result = sock_sendmsg(sock, msg, size);
697 set_fs(oldfs);
698 return result;
699}
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700700EXPORT_SYMBOL(kernel_sendmsg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700701
Eric Dumazet92f37fd2007-03-25 22:14:49 -0700702/*
703 * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP)
704 */
705void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk,
706 struct sk_buff *skb)
707{
Patrick Ohly20d49472009-02-12 05:03:38 +0000708 int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP);
Willem de Bruijnf24b9be2014-08-04 22:11:45 -0400709 struct scm_timestamping tss;
Patrick Ohly20d49472009-02-12 05:03:38 +0000710 int empty = 1;
711 struct skb_shared_hwtstamps *shhwtstamps =
712 skb_hwtstamps(skb);
Eric Dumazet92f37fd2007-03-25 22:14:49 -0700713
Patrick Ohly20d49472009-02-12 05:03:38 +0000714 /* Race occurred between timestamp enabling and packet
715 receiving. Fill in the current time for now. */
716 if (need_software_tstamp && skb->tstamp.tv64 == 0)
717 __net_timestamp(skb);
718
719 if (need_software_tstamp) {
720 if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) {
721 struct timeval tv;
722 skb_get_timestamp(skb, &tv);
723 put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMP,
724 sizeof(tv), &tv);
725 } else {
Willem de Bruijnf24b9be2014-08-04 22:11:45 -0400726 struct timespec ts;
727 skb_get_timestampns(skb, &ts);
Patrick Ohly20d49472009-02-12 05:03:38 +0000728 put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPNS,
Willem de Bruijnf24b9be2014-08-04 22:11:45 -0400729 sizeof(ts), &ts);
Patrick Ohly20d49472009-02-12 05:03:38 +0000730 }
Eric Dumazet92f37fd2007-03-25 22:14:49 -0700731 }
Patrick Ohly20d49472009-02-12 05:03:38 +0000732
Willem de Bruijnf24b9be2014-08-04 22:11:45 -0400733 memset(&tss, 0, sizeof(tss));
Willem de Bruijnc1991052014-09-03 12:01:18 -0400734 if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) &&
Willem de Bruijnf24b9be2014-08-04 22:11:45 -0400735 ktime_to_timespec_cond(skb->tstamp, tss.ts + 0))
Patrick Ohly20d49472009-02-12 05:03:38 +0000736 empty = 0;
Willem de Bruijn4d276eb2014-07-25 18:01:32 -0400737 if (shhwtstamps &&
Willem de Bruijnb9f40e22014-08-04 22:11:46 -0400738 (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) &&
Willem de Bruijnf24b9be2014-08-04 22:11:45 -0400739 ktime_to_timespec_cond(shhwtstamps->hwtstamp, tss.ts + 2))
Willem de Bruijn4d276eb2014-07-25 18:01:32 -0400740 empty = 0;
Patrick Ohly20d49472009-02-12 05:03:38 +0000741 if (!empty)
742 put_cmsg(msg, SOL_SOCKET,
Willem de Bruijnf24b9be2014-08-04 22:11:45 -0400743 SCM_TIMESTAMPING, sizeof(tss), &tss);
Eric Dumazet92f37fd2007-03-25 22:14:49 -0700744}
Arnaldo Carvalho de Melo7c81fd82007-03-10 00:39:35 -0300745EXPORT_SYMBOL_GPL(__sock_recv_timestamp);
746
Johannes Berg6e3e9392011-11-09 10:15:42 +0100747void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk,
748 struct sk_buff *skb)
749{
750 int ack;
751
752 if (!sock_flag(sk, SOCK_WIFI_STATUS))
753 return;
754 if (!skb->wifi_acked_valid)
755 return;
756
757 ack = skb->wifi_acked;
758
759 put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack);
760}
761EXPORT_SYMBOL_GPL(__sock_recv_wifi_status);
762
stephen hemminger11165f12010-10-18 14:27:29 +0000763static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk,
764 struct sk_buff *skb)
Neil Horman3b885782009-10-12 13:26:31 -0700765{
766 if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && skb->dropcount)
767 put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL,
768 sizeof(__u32), &skb->dropcount);
769}
770
Eric Dumazet767dd032010-04-28 19:14:43 +0000771void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk,
Neil Horman3b885782009-10-12 13:26:31 -0700772 struct sk_buff *skb)
773{
774 sock_recv_timestamp(msg, sk, skb);
775 sock_recv_drops(msg, sk, skb);
776}
Eric Dumazet767dd032010-04-28 19:14:43 +0000777EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops);
Neil Horman3b885782009-10-12 13:26:31 -0700778
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -0700779static inline int __sock_recvmsg_nosec(struct kiocb *iocb, struct socket *sock,
780 struct msghdr *msg, size_t size, int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700781{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700782 struct sock_iocb *si = kiocb_to_siocb(iocb);
783
784 si->sock = sock;
785 si->scm = NULL;
786 si->msg = msg;
787 si->size = size;
788 si->flags = flags;
789
Linus Torvalds1da177e2005-04-16 15:20:36 -0700790 return sock->ops->recvmsg(iocb, sock, msg, size, flags);
791}
792
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -0700793static inline int __sock_recvmsg(struct kiocb *iocb, struct socket *sock,
794 struct msghdr *msg, size_t size, int flags)
795{
796 int err = security_socket_recvmsg(sock, msg, size, flags);
797
798 return err ?: __sock_recvmsg_nosec(iocb, sock, msg, size, flags);
799}
800
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700801int sock_recvmsg(struct socket *sock, struct msghdr *msg,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700802 size_t size, int flags)
803{
804 struct kiocb iocb;
805 struct sock_iocb siocb;
806 int ret;
807
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700808 init_sync_kiocb(&iocb, NULL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700809 iocb.private = &siocb;
810 ret = __sock_recvmsg(&iocb, sock, msg, size, flags);
811 if (-EIOCBQUEUED == ret)
812 ret = wait_on_sync_kiocb(&iocb);
813 return ret;
814}
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700815EXPORT_SYMBOL(sock_recvmsg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700816
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -0700817static int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg,
818 size_t size, int flags)
819{
820 struct kiocb iocb;
821 struct sock_iocb siocb;
822 int ret;
823
824 init_sync_kiocb(&iocb, NULL);
825 iocb.private = &siocb;
826 ret = __sock_recvmsg_nosec(&iocb, sock, msg, size, flags);
827 if (-EIOCBQUEUED == ret)
828 ret = wait_on_sync_kiocb(&iocb);
829 return ret;
830}
831
Martin Lucinac1249c02010-12-10 00:04:05 +0000832/**
833 * kernel_recvmsg - Receive a message from a socket (kernel space)
834 * @sock: The socket to receive the message from
835 * @msg: Received message
836 * @vec: Input s/g array for message data
837 * @num: Size of input s/g array
838 * @size: Number of bytes to read
839 * @flags: Message flags (MSG_DONTWAIT, etc...)
840 *
841 * On return the msg structure contains the scatter/gather array passed in the
842 * vec argument. The array is modified so that it consists of the unfilled
843 * portion of the original array.
844 *
845 * The returned value is the total number of bytes received, or an error.
846 */
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700847int kernel_recvmsg(struct socket *sock, struct msghdr *msg,
848 struct kvec *vec, size_t num, size_t size, int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700849{
850 mm_segment_t oldfs = get_fs();
851 int result;
852
853 set_fs(KERNEL_DS);
854 /*
855 * the following is safe, since for compiler definitions of kvec and
856 * iovec are identical, yielding the same in-core layout and alignment
857 */
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700858 msg->msg_iov = (struct iovec *)vec, msg->msg_iovlen = num;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700859 result = sock_recvmsg(sock, msg, size, flags);
860 set_fs(oldfs);
861 return result;
862}
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700863EXPORT_SYMBOL(kernel_recvmsg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700864
Arnaldo Carvalho de Melo20380732005-08-16 02:18:02 -0300865static ssize_t sock_sendpage(struct file *file, struct page *page,
866 int offset, size_t size, loff_t *ppos, int more)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700867{
868 struct socket *sock;
869 int flags;
870
Eric Dumazetb69aee02005-09-06 14:42:45 -0700871 sock = file->private_data;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700872
Eric Dumazet35f9c092012-04-05 03:05:35 +0000873 flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0;
874 /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */
875 flags |= more;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700876
Linus Torvaldse6949582009-08-13 08:28:36 -0700877 return kernel_sendpage(sock, page, offset, size, flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700878}
879
Jens Axboe9c55e012007-11-06 23:30:13 -0800880static ssize_t sock_splice_read(struct file *file, loff_t *ppos,
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700881 struct pipe_inode_info *pipe, size_t len,
Jens Axboe9c55e012007-11-06 23:30:13 -0800882 unsigned int flags)
883{
884 struct socket *sock = file->private_data;
885
Rémi Denis-Courmont997b37d2008-02-15 02:35:45 -0800886 if (unlikely(!sock->ops->splice_read))
887 return -EINVAL;
888
Jens Axboe9c55e012007-11-06 23:30:13 -0800889 return sock->ops->splice_read(sock, ppos, pipe, len, flags);
890}
891
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800892static struct sock_iocb *alloc_sock_iocb(struct kiocb *iocb,
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700893 struct sock_iocb *siocb)
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800894{
Kent Overstreetd29c4452013-03-18 11:09:26 -0700895 if (!is_sync_kiocb(iocb))
896 BUG();
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800897
898 siocb->kiocb = iocb;
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800899 iocb->private = siocb;
900 return siocb;
901}
902
903static ssize_t do_sock_read(struct msghdr *msg, struct kiocb *iocb,
Badari Pulavarty027445c2006-09-30 23:28:46 -0700904 struct file *file, const struct iovec *iov,
905 unsigned long nr_segs)
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800906{
907 struct socket *sock = file->private_data;
908 size_t size = 0;
909 int i;
910
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700911 for (i = 0; i < nr_segs; i++)
912 size += iov[i].iov_len;
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800913
914 msg->msg_name = NULL;
915 msg->msg_namelen = 0;
916 msg->msg_control = NULL;
917 msg->msg_controllen = 0;
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700918 msg->msg_iov = (struct iovec *)iov;
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800919 msg->msg_iovlen = nr_segs;
920 msg->msg_flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0;
921
922 return __sock_recvmsg(iocb, sock, msg, size, msg->msg_flags);
923}
924
Badari Pulavarty027445c2006-09-30 23:28:46 -0700925static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov,
926 unsigned long nr_segs, loff_t pos)
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800927{
928 struct sock_iocb siocb, *x;
929
930 if (pos != 0)
931 return -ESPIPE;
Badari Pulavarty027445c2006-09-30 23:28:46 -0700932
Kent Overstreet73a70752013-05-09 15:03:42 -0700933 if (iocb->ki_nbytes == 0) /* Match SYS5 behaviour */
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800934 return 0;
935
Badari Pulavarty027445c2006-09-30 23:28:46 -0700936
937 x = alloc_sock_iocb(iocb, &siocb);
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800938 if (!x)
939 return -ENOMEM;
Badari Pulavarty027445c2006-09-30 23:28:46 -0700940 return do_sock_read(&x->async_msg, iocb, iocb->ki_filp, iov, nr_segs);
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800941}
942
943static ssize_t do_sock_write(struct msghdr *msg, struct kiocb *iocb,
Badari Pulavarty027445c2006-09-30 23:28:46 -0700944 struct file *file, const struct iovec *iov,
945 unsigned long nr_segs)
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800946{
947 struct socket *sock = file->private_data;
948 size_t size = 0;
949 int i;
950
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700951 for (i = 0; i < nr_segs; i++)
952 size += iov[i].iov_len;
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800953
954 msg->msg_name = NULL;
955 msg->msg_namelen = 0;
956 msg->msg_control = NULL;
957 msg->msg_controllen = 0;
Stephen Hemminger89bddce2006-09-01 00:19:31 -0700958 msg->msg_iov = (struct iovec *)iov;
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800959 msg->msg_iovlen = nr_segs;
960 msg->msg_flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0;
961 if (sock->type == SOCK_SEQPACKET)
962 msg->msg_flags |= MSG_EOR;
963
964 return __sock_sendmsg(iocb, sock, msg, size);
965}
966
Badari Pulavarty027445c2006-09-30 23:28:46 -0700967static ssize_t sock_aio_write(struct kiocb *iocb, const struct iovec *iov,
968 unsigned long nr_segs, loff_t pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700969{
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800970 struct sock_iocb siocb, *x;
971
972 if (pos != 0)
973 return -ESPIPE;
Badari Pulavarty027445c2006-09-30 23:28:46 -0700974
Badari Pulavarty027445c2006-09-30 23:28:46 -0700975 x = alloc_sock_iocb(iocb, &siocb);
Christoph Hellwigce1d4d32005-12-22 21:08:46 -0800976 if (!x)
977 return -ENOMEM;
978
Badari Pulavarty027445c2006-09-30 23:28:46 -0700979 return do_sock_write(&x->async_msg, iocb, iocb->ki_filp, iov, nr_segs);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700980}
981
Linus Torvalds1da177e2005-04-16 15:20:36 -0700982/*
983 * Atomic setting of ioctl hooks to avoid race
984 * with module unload.
985 */
986
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -0800987static DEFINE_MUTEX(br_ioctl_mutex);
Eric Dumazetc6d409c2010-06-03 20:03:40 -0700988static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700989
Eric W. Biederman881d9662007-09-17 11:56:21 -0700990void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700991{
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -0800992 mutex_lock(&br_ioctl_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700993 br_ioctl_hook = hook;
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -0800994 mutex_unlock(&br_ioctl_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700995}
996EXPORT_SYMBOL(brioctl_set);
997
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -0800998static DEFINE_MUTEX(vlan_ioctl_mutex);
Eric W. Biederman881d9662007-09-17 11:56:21 -0700999static int (*vlan_ioctl_hook) (struct net *, void __user *arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001000
Eric W. Biederman881d9662007-09-17 11:56:21 -07001001void vlan_ioctl_set(int (*hook) (struct net *, void __user *))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001002{
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -08001003 mutex_lock(&vlan_ioctl_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001004 vlan_ioctl_hook = hook;
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -08001005 mutex_unlock(&vlan_ioctl_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001006}
1007EXPORT_SYMBOL(vlan_ioctl_set);
1008
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -08001009static DEFINE_MUTEX(dlci_ioctl_mutex);
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001010static int (*dlci_ioctl_hook) (unsigned int, void __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001011
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001012void dlci_ioctl_set(int (*hook) (unsigned int, void __user *))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001013{
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -08001014 mutex_lock(&dlci_ioctl_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001015 dlci_ioctl_hook = hook;
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -08001016 mutex_unlock(&dlci_ioctl_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001017}
1018EXPORT_SYMBOL(dlci_ioctl_set);
1019
Arnd Bergmann6b960182009-11-06 23:10:54 -08001020static long sock_do_ioctl(struct net *net, struct socket *sock,
1021 unsigned int cmd, unsigned long arg)
1022{
1023 int err;
1024 void __user *argp = (void __user *)arg;
1025
1026 err = sock->ops->ioctl(sock, cmd, arg);
1027
1028 /*
1029 * If this ioctl is unknown try to hand it down
1030 * to the NIC driver.
1031 */
1032 if (err == -ENOIOCTLCMD)
1033 err = dev_ioctl(net, cmd, argp);
1034
1035 return err;
1036}
1037
Linus Torvalds1da177e2005-04-16 15:20:36 -07001038/*
1039 * With an ioctl, arg may well be a user mode pointer, but we don't know
1040 * what to do with it - that's up to the protocol still.
1041 */
1042
1043static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg)
1044{
1045 struct socket *sock;
Eric W. Biederman881d9662007-09-17 11:56:21 -07001046 struct sock *sk;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001047 void __user *argp = (void __user *)arg;
1048 int pid, err;
Eric W. Biederman881d9662007-09-17 11:56:21 -07001049 struct net *net;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001050
Eric Dumazetb69aee02005-09-06 14:42:45 -07001051 sock = file->private_data;
Eric W. Biederman881d9662007-09-17 11:56:21 -07001052 sk = sock->sk;
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +09001053 net = sock_net(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001054 if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) {
Eric W. Biederman881d9662007-09-17 11:56:21 -07001055 err = dev_ioctl(net, cmd, argp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001056 } else
Johannes Berg3d23e342009-09-29 23:27:28 +02001057#ifdef CONFIG_WEXT_CORE
Linus Torvalds1da177e2005-04-16 15:20:36 -07001058 if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) {
Eric W. Biederman881d9662007-09-17 11:56:21 -07001059 err = dev_ioctl(net, cmd, argp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001060 } else
Johannes Berg3d23e342009-09-29 23:27:28 +02001061#endif
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001062 switch (cmd) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001063 case FIOSETOWN:
1064 case SIOCSPGRP:
1065 err = -EFAULT;
1066 if (get_user(pid, (int __user *)argp))
1067 break;
Jeff Laytone0b93ed2014-08-22 11:27:32 -04001068 f_setown(sock->file, pid, 1);
1069 err = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001070 break;
1071 case FIOGETOWN:
1072 case SIOCGPGRP:
Eric W. Biederman609d7fa2006-10-02 02:17:15 -07001073 err = put_user(f_getown(sock->file),
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001074 (int __user *)argp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001075 break;
1076 case SIOCGIFBR:
1077 case SIOCSIFBR:
1078 case SIOCBRADDBR:
1079 case SIOCBRDELBR:
1080 err = -ENOPKG;
1081 if (!br_ioctl_hook)
1082 request_module("bridge");
1083
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -08001084 mutex_lock(&br_ioctl_mutex);
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001085 if (br_ioctl_hook)
Eric W. Biederman881d9662007-09-17 11:56:21 -07001086 err = br_ioctl_hook(net, cmd, argp);
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -08001087 mutex_unlock(&br_ioctl_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001088 break;
1089 case SIOCGIFVLAN:
1090 case SIOCSIFVLAN:
1091 err = -ENOPKG;
1092 if (!vlan_ioctl_hook)
1093 request_module("8021q");
1094
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -08001095 mutex_lock(&vlan_ioctl_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001096 if (vlan_ioctl_hook)
Eric W. Biederman881d9662007-09-17 11:56:21 -07001097 err = vlan_ioctl_hook(net, argp);
Arjan van de Ven4a3e2f72006-03-20 22:33:17 -08001098 mutex_unlock(&vlan_ioctl_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001099 break;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001100 case SIOCADDDLCI:
1101 case SIOCDELDLCI:
1102 err = -ENOPKG;
1103 if (!dlci_ioctl_hook)
1104 request_module("dlci");
1105
Pavel Emelyanov7512cbf2008-03-21 15:58:52 -07001106 mutex_lock(&dlci_ioctl_mutex);
1107 if (dlci_ioctl_hook)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001108 err = dlci_ioctl_hook(cmd, argp);
Pavel Emelyanov7512cbf2008-03-21 15:58:52 -07001109 mutex_unlock(&dlci_ioctl_mutex);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001110 break;
1111 default:
Arnd Bergmann6b960182009-11-06 23:10:54 -08001112 err = sock_do_ioctl(net, sock, cmd, arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001113 break;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001114 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001115 return err;
1116}
1117
1118int sock_create_lite(int family, int type, int protocol, struct socket **res)
1119{
1120 int err;
1121 struct socket *sock = NULL;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001122
Linus Torvalds1da177e2005-04-16 15:20:36 -07001123 err = security_socket_create(family, type, protocol, 1);
1124 if (err)
1125 goto out;
1126
1127 sock = sock_alloc();
1128 if (!sock) {
1129 err = -ENOMEM;
1130 goto out;
1131 }
1132
Linus Torvalds1da177e2005-04-16 15:20:36 -07001133 sock->type = type;
Venkat Yekkirala7420ed22006-08-04 23:17:57 -07001134 err = security_socket_post_create(sock, family, type, protocol, 1);
1135 if (err)
1136 goto out_release;
1137
Linus Torvalds1da177e2005-04-16 15:20:36 -07001138out:
1139 *res = sock;
1140 return err;
Venkat Yekkirala7420ed22006-08-04 23:17:57 -07001141out_release:
1142 sock_release(sock);
1143 sock = NULL;
1144 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001145}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07001146EXPORT_SYMBOL(sock_create_lite);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001147
1148/* No kernel lock held - perfect */
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001149static unsigned int sock_poll(struct file *file, poll_table *wait)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001150{
Eliezer Tamircbf55002013-07-08 16:20:34 +03001151 unsigned int busy_flag = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001152 struct socket *sock;
1153
1154 /*
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001155 * We can't return errors to poll, so it's either yes or no.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001156 */
Eric Dumazetb69aee02005-09-06 14:42:45 -07001157 sock = file->private_data;
Eliezer Tamir2d48d672013-06-24 10:28:03 +03001158
Eliezer Tamircbf55002013-07-08 16:20:34 +03001159 if (sk_can_busy_loop(sock->sk)) {
Eliezer Tamir2d48d672013-06-24 10:28:03 +03001160 /* this socket can poll_ll so tell the system call */
Eliezer Tamircbf55002013-07-08 16:20:34 +03001161 busy_flag = POLL_BUSY_LOOP;
Eliezer Tamir2d48d672013-06-24 10:28:03 +03001162
1163 /* once, only if requested by syscall */
Eliezer Tamircbf55002013-07-08 16:20:34 +03001164 if (wait && (wait->_key & POLL_BUSY_LOOP))
1165 sk_busy_loop(sock->sk, 1);
Eliezer Tamir2d48d672013-06-24 10:28:03 +03001166 }
1167
Eliezer Tamircbf55002013-07-08 16:20:34 +03001168 return busy_flag | sock->ops->poll(file, sock, wait);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001169}
1170
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001171static int sock_mmap(struct file *file, struct vm_area_struct *vma)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001172{
Eric Dumazetb69aee02005-09-06 14:42:45 -07001173 struct socket *sock = file->private_data;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001174
1175 return sock->ops->mmap(file, sock, vma);
1176}
1177
Arnaldo Carvalho de Melo20380732005-08-16 02:18:02 -03001178static int sock_close(struct inode *inode, struct file *filp)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001179{
Linus Torvalds1da177e2005-04-16 15:20:36 -07001180 sock_release(SOCKET_I(inode));
1181 return 0;
1182}
1183
1184/*
1185 * Update the socket async list
1186 *
1187 * Fasync_list locking strategy.
1188 *
1189 * 1. fasync_list is modified only under process context socket lock
1190 * i.e. under semaphore.
1191 * 2. fasync_list is used under read_lock(&sk->sk_callback_lock)
Eric Dumazet989a2972010-04-14 09:55:35 +00001192 * or under socket lock
Linus Torvalds1da177e2005-04-16 15:20:36 -07001193 */
1194
1195static int sock_fasync(int fd, struct file *filp, int on)
1196{
Eric Dumazet989a2972010-04-14 09:55:35 +00001197 struct socket *sock = filp->private_data;
1198 struct sock *sk = sock->sk;
Eric Dumazeteaefd1102011-02-18 03:26:36 +00001199 struct socket_wq *wq;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001200
Eric Dumazet989a2972010-04-14 09:55:35 +00001201 if (sk == NULL)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001202 return -EINVAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001203
1204 lock_sock(sk);
Eric Dumazeteaefd1102011-02-18 03:26:36 +00001205 wq = rcu_dereference_protected(sock->wq, sock_owned_by_user(sk));
1206 fasync_helper(fd, filp, on, &wq->fasync_list);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001207
Eric Dumazeteaefd1102011-02-18 03:26:36 +00001208 if (!wq->fasync_list)
Eric Dumazet989a2972010-04-14 09:55:35 +00001209 sock_reset_flag(sk, SOCK_FASYNC);
Jonathan Corbet76398422009-02-01 14:26:59 -07001210 else
Eric Dumazetbcdce712009-10-06 17:28:29 -07001211 sock_set_flag(sk, SOCK_FASYNC);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001212
Eric Dumazet989a2972010-04-14 09:55:35 +00001213 release_sock(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001214 return 0;
1215}
1216
Eric Dumazet43815482010-04-29 11:01:49 +00001217/* This function may be called only under socket lock or callback_lock or rcu_lock */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001218
1219int sock_wake_async(struct socket *sock, int how, int band)
1220{
Eric Dumazet43815482010-04-29 11:01:49 +00001221 struct socket_wq *wq;
1222
1223 if (!sock)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001224 return -1;
Eric Dumazet43815482010-04-29 11:01:49 +00001225 rcu_read_lock();
1226 wq = rcu_dereference(sock->wq);
1227 if (!wq || !wq->fasync_list) {
1228 rcu_read_unlock();
1229 return -1;
1230 }
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001231 switch (how) {
Pavel Emelyanov8d8ad9d2007-11-26 20:10:50 +08001232 case SOCK_WAKE_WAITD:
Linus Torvalds1da177e2005-04-16 15:20:36 -07001233 if (test_bit(SOCK_ASYNC_WAITDATA, &sock->flags))
1234 break;
1235 goto call_kill;
Pavel Emelyanov8d8ad9d2007-11-26 20:10:50 +08001236 case SOCK_WAKE_SPACE:
Linus Torvalds1da177e2005-04-16 15:20:36 -07001237 if (!test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags))
1238 break;
1239 /* fall through */
Pavel Emelyanov8d8ad9d2007-11-26 20:10:50 +08001240 case SOCK_WAKE_IO:
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001241call_kill:
Eric Dumazet43815482010-04-29 11:01:49 +00001242 kill_fasync(&wq->fasync_list, SIGIO, band);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001243 break;
Pavel Emelyanov8d8ad9d2007-11-26 20:10:50 +08001244 case SOCK_WAKE_URG:
Eric Dumazet43815482010-04-29 11:01:49 +00001245 kill_fasync(&wq->fasync_list, SIGURG, band);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001246 }
Eric Dumazet43815482010-04-29 11:01:49 +00001247 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -07001248 return 0;
1249}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07001250EXPORT_SYMBOL(sock_wake_async);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001251
Pavel Emelyanov721db932010-09-29 16:06:32 +04001252int __sock_create(struct net *net, int family, int type, int protocol,
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001253 struct socket **res, int kern)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001254{
1255 int err;
1256 struct socket *sock;
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001257 const struct net_proto_family *pf;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001258
1259 /*
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001260 * Check protocol is in range
Linus Torvalds1da177e2005-04-16 15:20:36 -07001261 */
1262 if (family < 0 || family >= NPROTO)
1263 return -EAFNOSUPPORT;
1264 if (type < 0 || type >= SOCK_MAX)
1265 return -EINVAL;
1266
1267 /* Compatibility.
1268
1269 This uglymoron is moved from INET layer to here to avoid
1270 deadlock in module load.
1271 */
1272 if (family == PF_INET && type == SOCK_PACKET) {
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001273 static int warned;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001274 if (!warned) {
1275 warned = 1;
Yang Yingliang3410f222014-02-12 17:09:55 +08001276 pr_info("%s uses obsolete (PF_INET,SOCK_PACKET)\n",
1277 current->comm);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001278 }
1279 family = PF_PACKET;
1280 }
1281
1282 err = security_socket_create(family, type, protocol, kern);
1283 if (err)
1284 return err;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001285
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001286 /*
1287 * Allocate the socket and allow the family to set things up. if
1288 * the protocol is 0, the family is instructed to select an appropriate
1289 * default.
1290 */
1291 sock = sock_alloc();
1292 if (!sock) {
Joe Perchese87cc472012-05-13 21:56:26 +00001293 net_warn_ratelimited("socket: no more sockets\n");
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001294 return -ENFILE; /* Not exactly a match, but its the
1295 closest posix thing */
1296 }
1297
1298 sock->type = type;
1299
Johannes Berg95a5afc2008-10-16 15:24:51 -07001300#ifdef CONFIG_MODULES
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001301 /* Attempt to load a protocol module if the find failed.
1302 *
1303 * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user
Linus Torvalds1da177e2005-04-16 15:20:36 -07001304 * requested real, full-featured networking support upon configuration.
1305 * Otherwise module support will break!
1306 */
Eric Dumazet190683a2010-11-10 10:50:44 +00001307 if (rcu_access_pointer(net_families[family]) == NULL)
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001308 request_module("net-pf-%d", family);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001309#endif
1310
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001311 rcu_read_lock();
1312 pf = rcu_dereference(net_families[family]);
1313 err = -EAFNOSUPPORT;
1314 if (!pf)
1315 goto out_release;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001316
1317 /*
1318 * We will call the ->create function, that possibly is in a loadable
1319 * module, so we have to bump that loadable module refcnt first.
1320 */
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001321 if (!try_module_get(pf->owner))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001322 goto out_release;
1323
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001324 /* Now protected by module ref count */
1325 rcu_read_unlock();
1326
Eric Paris3f378b682009-11-05 22:18:14 -08001327 err = pf->create(net, sock, protocol, kern);
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001328 if (err < 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001329 goto out_module_put;
Frank Filza79af592005-09-27 15:23:38 -07001330
Linus Torvalds1da177e2005-04-16 15:20:36 -07001331 /*
1332 * Now to bump the refcnt of the [loadable] module that owns this
1333 * socket at sock_release time we decrement its refcnt.
1334 */
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001335 if (!try_module_get(sock->ops->owner))
1336 goto out_module_busy;
1337
Linus Torvalds1da177e2005-04-16 15:20:36 -07001338 /*
1339 * Now that we're done with the ->create function, the [loadable]
1340 * module can have its refcnt decremented
1341 */
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001342 module_put(pf->owner);
Venkat Yekkirala7420ed22006-08-04 23:17:57 -07001343 err = security_socket_post_create(sock, family, type, protocol, kern);
1344 if (err)
Herbert Xu3b185522007-08-15 14:46:02 -07001345 goto out_sock_release;
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001346 *res = sock;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001347
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001348 return 0;
1349
1350out_module_busy:
1351 err = -EAFNOSUPPORT;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001352out_module_put:
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001353 sock->ops = NULL;
1354 module_put(pf->owner);
1355out_sock_release:
Linus Torvalds1da177e2005-04-16 15:20:36 -07001356 sock_release(sock);
Stephen Hemminger55737fd2006-09-01 00:23:39 -07001357 return err;
1358
1359out_release:
1360 rcu_read_unlock();
1361 goto out_sock_release;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001362}
Pavel Emelyanov721db932010-09-29 16:06:32 +04001363EXPORT_SYMBOL(__sock_create);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001364
1365int sock_create(int family, int type, int protocol, struct socket **res)
1366{
Eric W. Biederman1b8d7ae2007-10-08 23:24:22 -07001367 return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001368}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07001369EXPORT_SYMBOL(sock_create);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001370
1371int sock_create_kern(int family, int type, int protocol, struct socket **res)
1372{
Eric W. Biederman1b8d7ae2007-10-08 23:24:22 -07001373 return __sock_create(&init_net, family, type, protocol, res, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001374}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07001375EXPORT_SYMBOL(sock_create_kern);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001376
Heiko Carstens3e0fa652009-01-14 14:14:24 +01001377SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001378{
1379 int retval;
1380 struct socket *sock;
Ulrich Dreppera677a032008-07-23 21:29:17 -07001381 int flags;
1382
Ulrich Dreppere38b36f2008-07-23 21:29:42 -07001383 /* Check the SOCK_* constants for consistency. */
1384 BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC);
1385 BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK);
1386 BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK);
1387 BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK);
1388
Ulrich Dreppera677a032008-07-23 21:29:17 -07001389 flags = type & ~SOCK_TYPE_MASK;
Ulrich Drepper77d27202008-07-23 21:29:35 -07001390 if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK))
Ulrich Dreppera677a032008-07-23 21:29:17 -07001391 return -EINVAL;
1392 type &= SOCK_TYPE_MASK;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001393
Ulrich Drepperaaca0bd2008-07-23 21:29:20 -07001394 if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK))
1395 flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK;
1396
Linus Torvalds1da177e2005-04-16 15:20:36 -07001397 retval = sock_create(family, type, protocol, &sock);
1398 if (retval < 0)
1399 goto out;
1400
Ulrich Drepper77d27202008-07-23 21:29:35 -07001401 retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001402 if (retval < 0)
1403 goto out_release;
1404
1405out:
1406 /* It may be already another descriptor 8) Not kernel problem. */
1407 return retval;
1408
1409out_release:
1410 sock_release(sock);
1411 return retval;
1412}
1413
1414/*
1415 * Create a pair of connected sockets.
1416 */
1417
Heiko Carstens3e0fa652009-01-14 14:14:24 +01001418SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol,
1419 int __user *, usockvec)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001420{
1421 struct socket *sock1, *sock2;
1422 int fd1, fd2, err;
Al Virodb349502007-02-07 01:48:00 -05001423 struct file *newfile1, *newfile2;
Ulrich Dreppera677a032008-07-23 21:29:17 -07001424 int flags;
1425
1426 flags = type & ~SOCK_TYPE_MASK;
Ulrich Drepper77d27202008-07-23 21:29:35 -07001427 if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK))
Ulrich Dreppera677a032008-07-23 21:29:17 -07001428 return -EINVAL;
1429 type &= SOCK_TYPE_MASK;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001430
Ulrich Drepperaaca0bd2008-07-23 21:29:20 -07001431 if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK))
1432 flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK;
1433
Linus Torvalds1da177e2005-04-16 15:20:36 -07001434 /*
1435 * Obtain the first socket and check if the underlying protocol
1436 * supports the socketpair call.
1437 */
1438
1439 err = sock_create(family, type, protocol, &sock1);
1440 if (err < 0)
1441 goto out;
1442
1443 err = sock_create(family, type, protocol, &sock2);
1444 if (err < 0)
1445 goto out_release_1;
1446
1447 err = sock1->ops->socketpair(sock1, sock2);
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001448 if (err < 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001449 goto out_release_both;
1450
Al Viro28407632012-08-17 23:54:15 -04001451 fd1 = get_unused_fd_flags(flags);
David S. Millerbf3c23d2007-10-29 21:54:02 -07001452 if (unlikely(fd1 < 0)) {
1453 err = fd1;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001454 goto out_release_both;
David S. Millerbf3c23d2007-10-29 21:54:02 -07001455 }
Yann Droneaudd73aa282013-12-09 22:42:20 +01001456
Al Viro28407632012-08-17 23:54:15 -04001457 fd2 = get_unused_fd_flags(flags);
Al Viro198de4d2009-08-05 19:29:23 +04001458 if (unlikely(fd2 < 0)) {
1459 err = fd2;
Yann Droneaudd73aa282013-12-09 22:42:20 +01001460 goto out_put_unused_1;
Al Viro28407632012-08-17 23:54:15 -04001461 }
1462
Linus Torvaldsaab174f2012-10-02 20:25:04 -07001463 newfile1 = sock_alloc_file(sock1, flags, NULL);
Al Viro28407632012-08-17 23:54:15 -04001464 if (unlikely(IS_ERR(newfile1))) {
1465 err = PTR_ERR(newfile1);
Yann Droneaudd73aa282013-12-09 22:42:20 +01001466 goto out_put_unused_both;
Al Viro28407632012-08-17 23:54:15 -04001467 }
1468
Linus Torvaldsaab174f2012-10-02 20:25:04 -07001469 newfile2 = sock_alloc_file(sock2, flags, NULL);
Al Viro28407632012-08-17 23:54:15 -04001470 if (IS_ERR(newfile2)) {
1471 err = PTR_ERR(newfile2);
Yann Droneaudd73aa282013-12-09 22:42:20 +01001472 goto out_fput_1;
Al Virodb349502007-02-07 01:48:00 -05001473 }
1474
Yann Droneaudd73aa282013-12-09 22:42:20 +01001475 err = put_user(fd1, &usockvec[0]);
1476 if (err)
1477 goto out_fput_both;
1478
1479 err = put_user(fd2, &usockvec[1]);
1480 if (err)
1481 goto out_fput_both;
1482
Al Viro157cf642008-12-14 04:57:47 -05001483 audit_fd_pair(fd1, fd2);
Yann Droneaudd73aa282013-12-09 22:42:20 +01001484
Al Virodb349502007-02-07 01:48:00 -05001485 fd_install(fd1, newfile1);
1486 fd_install(fd2, newfile2);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001487 /* fd1 and fd2 may be already another descriptors.
1488 * Not kernel problem.
1489 */
1490
Yann Droneaudd73aa282013-12-09 22:42:20 +01001491 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001492
Yann Droneaudd73aa282013-12-09 22:42:20 +01001493out_fput_both:
1494 fput(newfile2);
1495 fput(newfile1);
1496 put_unused_fd(fd2);
1497 put_unused_fd(fd1);
1498 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001499
Yann Droneaudd73aa282013-12-09 22:42:20 +01001500out_fput_1:
1501 fput(newfile1);
1502 put_unused_fd(fd2);
1503 put_unused_fd(fd1);
1504 sock_release(sock2);
1505 goto out;
1506
1507out_put_unused_both:
1508 put_unused_fd(fd2);
1509out_put_unused_1:
1510 put_unused_fd(fd1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001511out_release_both:
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001512 sock_release(sock2);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001513out_release_1:
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001514 sock_release(sock1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001515out:
1516 return err;
1517}
1518
Linus Torvalds1da177e2005-04-16 15:20:36 -07001519/*
1520 * Bind a name to a socket. Nothing much to do here since it's
1521 * the protocol's responsibility to handle the local address.
1522 *
1523 * We move the socket address to kernel space before we call
1524 * the protocol layer (having also checked the address is ok).
1525 */
1526
Heiko Carstens20f37032009-01-14 14:14:23 +01001527SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001528{
1529 struct socket *sock;
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001530 struct sockaddr_storage address;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001531 int err, fput_needed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001532
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001533 sock = sockfd_lookup_light(fd, &err, &fput_needed);
Stephen Hemmingere71a4782007-04-10 20:10:33 -07001534 if (sock) {
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00001535 err = move_addr_to_kernel(umyaddr, addrlen, &address);
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001536 if (err >= 0) {
1537 err = security_socket_bind(sock,
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001538 (struct sockaddr *)&address,
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001539 addrlen);
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001540 if (!err)
1541 err = sock->ops->bind(sock,
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001542 (struct sockaddr *)
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001543 &address, addrlen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001544 }
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001545 fput_light(sock->file, fput_needed);
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001546 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001547 return err;
1548}
1549
Linus Torvalds1da177e2005-04-16 15:20:36 -07001550/*
1551 * Perform a listen. Basically, we allow the protocol to do anything
1552 * necessary for a listen, and if that works, we mark the socket as
1553 * ready for listening.
1554 */
1555
Heiko Carstens3e0fa652009-01-14 14:14:24 +01001556SYSCALL_DEFINE2(listen, int, fd, int, backlog)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001557{
1558 struct socket *sock;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001559 int err, fput_needed;
Pavel Emelyanovb8e1f9b2007-12-08 00:12:33 -08001560 int somaxconn;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001561
1562 sock = sockfd_lookup_light(fd, &err, &fput_needed);
1563 if (sock) {
Pavel Emelyanov8efa6e92008-03-31 19:41:14 -07001564 somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn;
Eric Dumazet95c96172012-04-15 05:58:06 +00001565 if ((unsigned int)backlog > somaxconn)
Pavel Emelyanovb8e1f9b2007-12-08 00:12:33 -08001566 backlog = somaxconn;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001567
1568 err = security_socket_listen(sock, backlog);
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001569 if (!err)
1570 err = sock->ops->listen(sock, backlog);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001571
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001572 fput_light(sock->file, fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001573 }
1574 return err;
1575}
1576
Linus Torvalds1da177e2005-04-16 15:20:36 -07001577/*
1578 * For accept, we attempt to create a new socket, set up the link
1579 * with the client, wake up the client, then return the new
1580 * connected fd. We collect the address of the connector in kernel
1581 * space and move it to user at the very end. This is unclean because
1582 * we open the socket then return an error.
1583 *
1584 * 1003.1g adds the ability to recvmsg() to query connection pending
1585 * status to recvmsg. We need to add that support in a way thats
1586 * clean when we restucture accept also.
1587 */
1588
Heiko Carstens20f37032009-01-14 14:14:23 +01001589SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr,
1590 int __user *, upeer_addrlen, int, flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001591{
1592 struct socket *sock, *newsock;
David S. Miller39d8c1b2006-03-20 17:13:49 -08001593 struct file *newfile;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001594 int err, len, newfd, fput_needed;
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001595 struct sockaddr_storage address;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001596
Ulrich Drepper77d27202008-07-23 21:29:35 -07001597 if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK))
Ulrich Drepperaaca0bd2008-07-23 21:29:20 -07001598 return -EINVAL;
1599
1600 if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK))
1601 flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK;
1602
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001603 sock = sockfd_lookup_light(fd, &err, &fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001604 if (!sock)
1605 goto out;
1606
1607 err = -ENFILE;
Eric Dumazetc6d409c2010-06-03 20:03:40 -07001608 newsock = sock_alloc();
1609 if (!newsock)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001610 goto out_put;
1611
1612 newsock->type = sock->type;
1613 newsock->ops = sock->ops;
1614
Linus Torvalds1da177e2005-04-16 15:20:36 -07001615 /*
1616 * We don't need try_module_get here, as the listening socket (sock)
1617 * has the protocol module (sock->ops->owner) held.
1618 */
1619 __module_get(newsock->ops->owner);
1620
Al Viro28407632012-08-17 23:54:15 -04001621 newfd = get_unused_fd_flags(flags);
David S. Miller39d8c1b2006-03-20 17:13:49 -08001622 if (unlikely(newfd < 0)) {
1623 err = newfd;
David S. Miller9a1875e2006-04-01 12:48:36 -08001624 sock_release(newsock);
1625 goto out_put;
David S. Miller39d8c1b2006-03-20 17:13:49 -08001626 }
Linus Torvaldsaab174f2012-10-02 20:25:04 -07001627 newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name);
Al Viro28407632012-08-17 23:54:15 -04001628 if (unlikely(IS_ERR(newfile))) {
1629 err = PTR_ERR(newfile);
1630 put_unused_fd(newfd);
1631 sock_release(newsock);
1632 goto out_put;
1633 }
David S. Miller39d8c1b2006-03-20 17:13:49 -08001634
Frank Filza79af592005-09-27 15:23:38 -07001635 err = security_socket_accept(sock, newsock);
1636 if (err)
David S. Miller39d8c1b2006-03-20 17:13:49 -08001637 goto out_fd;
Frank Filza79af592005-09-27 15:23:38 -07001638
Linus Torvalds1da177e2005-04-16 15:20:36 -07001639 err = sock->ops->accept(sock, newsock, sock->file->f_flags);
1640 if (err < 0)
David S. Miller39d8c1b2006-03-20 17:13:49 -08001641 goto out_fd;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001642
1643 if (upeer_sockaddr) {
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001644 if (newsock->ops->getname(newsock, (struct sockaddr *)&address,
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001645 &len, 2) < 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001646 err = -ECONNABORTED;
David S. Miller39d8c1b2006-03-20 17:13:49 -08001647 goto out_fd;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001648 }
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00001649 err = move_addr_to_user(&address,
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001650 len, upeer_sockaddr, upeer_addrlen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001651 if (err < 0)
David S. Miller39d8c1b2006-03-20 17:13:49 -08001652 goto out_fd;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001653 }
1654
1655 /* File flags are not inherited via accept() unlike another OSes. */
1656
David S. Miller39d8c1b2006-03-20 17:13:49 -08001657 fd_install(newfd, newfile);
1658 err = newfd;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001659
Linus Torvalds1da177e2005-04-16 15:20:36 -07001660out_put:
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001661 fput_light(sock->file, fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001662out:
1663 return err;
David S. Miller39d8c1b2006-03-20 17:13:49 -08001664out_fd:
David S. Miller9606a212006-04-01 01:00:14 -08001665 fput(newfile);
David S. Miller39d8c1b2006-03-20 17:13:49 -08001666 put_unused_fd(newfd);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001667 goto out_put;
1668}
1669
Heiko Carstens20f37032009-01-14 14:14:23 +01001670SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr,
1671 int __user *, upeer_addrlen)
Ulrich Drepperaaca0bd2008-07-23 21:29:20 -07001672{
Ulrich Drepperde11def2008-11-19 15:36:14 -08001673 return sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0);
Ulrich Drepperaaca0bd2008-07-23 21:29:20 -07001674}
1675
Linus Torvalds1da177e2005-04-16 15:20:36 -07001676/*
1677 * Attempt to connect to a socket with the server address. The address
1678 * is in user space so we verify it is OK and move it to kernel space.
1679 *
1680 * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to
1681 * break bindings
1682 *
1683 * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and
1684 * other SEQPACKET protocols that take time to connect() as it doesn't
1685 * include the -EINPROGRESS status for such sockets.
1686 */
1687
Heiko Carstens20f37032009-01-14 14:14:23 +01001688SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr,
1689 int, addrlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001690{
1691 struct socket *sock;
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001692 struct sockaddr_storage address;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001693 int err, fput_needed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001694
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001695 sock = sockfd_lookup_light(fd, &err, &fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001696 if (!sock)
1697 goto out;
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00001698 err = move_addr_to_kernel(uservaddr, addrlen, &address);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001699 if (err < 0)
1700 goto out_put;
1701
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001702 err =
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001703 security_socket_connect(sock, (struct sockaddr *)&address, addrlen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001704 if (err)
1705 goto out_put;
1706
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001707 err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001708 sock->file->f_flags);
1709out_put:
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001710 fput_light(sock->file, fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001711out:
1712 return err;
1713}
1714
1715/*
1716 * Get the local address ('name') of a socket object. Move the obtained
1717 * name to user space.
1718 */
1719
Heiko Carstens20f37032009-01-14 14:14:23 +01001720SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr,
1721 int __user *, usockaddr_len)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001722{
1723 struct socket *sock;
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001724 struct sockaddr_storage address;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001725 int len, err, fput_needed;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001726
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001727 sock = sockfd_lookup_light(fd, &err, &fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001728 if (!sock)
1729 goto out;
1730
1731 err = security_socket_getsockname(sock);
1732 if (err)
1733 goto out_put;
1734
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001735 err = sock->ops->getname(sock, (struct sockaddr *)&address, &len, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001736 if (err)
1737 goto out_put;
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00001738 err = move_addr_to_user(&address, len, usockaddr, usockaddr_len);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001739
1740out_put:
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001741 fput_light(sock->file, fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001742out:
1743 return err;
1744}
1745
1746/*
1747 * Get the remote address ('name') of a socket object. Move the obtained
1748 * name to user space.
1749 */
1750
Heiko Carstens20f37032009-01-14 14:14:23 +01001751SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr,
1752 int __user *, usockaddr_len)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001753{
1754 struct socket *sock;
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001755 struct sockaddr_storage address;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001756 int len, err, fput_needed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001757
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001758 sock = sockfd_lookup_light(fd, &err, &fput_needed);
1759 if (sock != NULL) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001760 err = security_socket_getpeername(sock);
1761 if (err) {
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001762 fput_light(sock->file, fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001763 return err;
1764 }
1765
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001766 err =
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001767 sock->ops->getname(sock, (struct sockaddr *)&address, &len,
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001768 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001769 if (!err)
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00001770 err = move_addr_to_user(&address, len, usockaddr,
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001771 usockaddr_len);
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001772 fput_light(sock->file, fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001773 }
1774 return err;
1775}
1776
1777/*
1778 * Send a datagram to a given address. We move the address into kernel
1779 * space and check the user space data area is readable before invoking
1780 * the protocol.
1781 */
1782
Heiko Carstens3e0fa652009-01-14 14:14:24 +01001783SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len,
Eric Dumazet95c96172012-04-15 05:58:06 +00001784 unsigned int, flags, struct sockaddr __user *, addr,
Heiko Carstens3e0fa652009-01-14 14:14:24 +01001785 int, addr_len)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001786{
1787 struct socket *sock;
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001788 struct sockaddr_storage address;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001789 int err;
1790 struct msghdr msg;
1791 struct iovec iov;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001792 int fput_needed;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001793
Linus Torvalds253eacc2010-10-30 16:43:10 -07001794 if (len > INT_MAX)
1795 len = INT_MAX;
Pavel Emelyanovde0fa952007-11-14 16:01:43 -08001796 sock = sockfd_lookup_light(fd, &err, &fput_needed);
1797 if (!sock)
David S. Miller4387ff752007-02-08 15:06:08 -08001798 goto out;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001799
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001800 iov.iov_base = buff;
1801 iov.iov_len = len;
1802 msg.msg_name = NULL;
1803 msg.msg_iov = &iov;
1804 msg.msg_iovlen = 1;
1805 msg.msg_control = NULL;
1806 msg.msg_controllen = 0;
1807 msg.msg_namelen = 0;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001808 if (addr) {
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00001809 err = move_addr_to_kernel(addr, addr_len, &address);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001810 if (err < 0)
1811 goto out_put;
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001812 msg.msg_name = (struct sockaddr *)&address;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001813 msg.msg_namelen = addr_len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001814 }
1815 if (sock->file->f_flags & O_NONBLOCK)
1816 flags |= MSG_DONTWAIT;
1817 msg.msg_flags = flags;
1818 err = sock_sendmsg(sock, &msg, len);
1819
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001820out_put:
Pavel Emelyanovde0fa952007-11-14 16:01:43 -08001821 fput_light(sock->file, fput_needed);
David S. Miller4387ff752007-02-08 15:06:08 -08001822out:
Linus Torvalds1da177e2005-04-16 15:20:36 -07001823 return err;
1824}
1825
1826/*
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001827 * Send a datagram down a socket.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001828 */
1829
Heiko Carstens3e0fa652009-01-14 14:14:24 +01001830SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len,
Eric Dumazet95c96172012-04-15 05:58:06 +00001831 unsigned int, flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001832{
1833 return sys_sendto(fd, buff, len, flags, NULL, 0);
1834}
1835
1836/*
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001837 * Receive a frame from the socket and optionally record the address of the
Linus Torvalds1da177e2005-04-16 15:20:36 -07001838 * sender. We verify the buffers are writable and if needed move the
1839 * sender address from kernel to user space.
1840 */
1841
Heiko Carstens3e0fa652009-01-14 14:14:24 +01001842SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size,
Eric Dumazet95c96172012-04-15 05:58:06 +00001843 unsigned int, flags, struct sockaddr __user *, addr,
Heiko Carstens3e0fa652009-01-14 14:14:24 +01001844 int __user *, addr_len)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001845{
1846 struct socket *sock;
1847 struct iovec iov;
1848 struct msghdr msg;
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001849 struct sockaddr_storage address;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001850 int err, err2;
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001851 int fput_needed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001852
Linus Torvalds253eacc2010-10-30 16:43:10 -07001853 if (size > INT_MAX)
1854 size = INT_MAX;
Pavel Emelyanovde0fa952007-11-14 16:01:43 -08001855 sock = sockfd_lookup_light(fd, &err, &fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001856 if (!sock)
Pavel Emelyanovde0fa952007-11-14 16:01:43 -08001857 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001858
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001859 msg.msg_control = NULL;
1860 msg.msg_controllen = 0;
1861 msg.msg_iovlen = 1;
1862 msg.msg_iov = &iov;
1863 iov.iov_len = size;
1864 iov.iov_base = ubuf;
Hannes Frederic Sowaf3d33422013-11-21 03:14:22 +01001865 /* Save some cycles and don't copy the address if not needed */
1866 msg.msg_name = addr ? (struct sockaddr *)&address : NULL;
1867 /* We assume all kernel code knows the size of sockaddr_storage */
1868 msg.msg_namelen = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001869 if (sock->file->f_flags & O_NONBLOCK)
1870 flags |= MSG_DONTWAIT;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001871 err = sock_recvmsg(sock, &msg, size, flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001872
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001873 if (err >= 0 && addr != NULL) {
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00001874 err2 = move_addr_to_user(&address,
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07001875 msg.msg_namelen, addr, addr_len);
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001876 if (err2 < 0)
1877 err = err2;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001878 }
Pavel Emelyanovde0fa952007-11-14 16:01:43 -08001879
1880 fput_light(sock->file, fput_needed);
David S. Miller4387ff752007-02-08 15:06:08 -08001881out:
Linus Torvalds1da177e2005-04-16 15:20:36 -07001882 return err;
1883}
1884
1885/*
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001886 * Receive a datagram from a socket.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001887 */
1888
Jan Glauberb7c0ddf2014-04-16 09:32:48 +02001889SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size,
1890 unsigned int, flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001891{
1892 return sys_recvfrom(fd, ubuf, size, flags, NULL, NULL);
1893}
1894
1895/*
1896 * Set a socket option. Because we don't know the option lengths we have
1897 * to pass the user mode parameter for the protocols to sort out.
1898 */
1899
Heiko Carstens20f37032009-01-14 14:14:23 +01001900SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname,
1901 char __user *, optval, int, optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001902{
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001903 int err, fput_needed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001904 struct socket *sock;
1905
1906 if (optlen < 0)
1907 return -EINVAL;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001908
1909 sock = sockfd_lookup_light(fd, &err, &fput_needed);
1910 if (sock != NULL) {
1911 err = security_socket_setsockopt(sock, level, optname);
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001912 if (err)
1913 goto out_put;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001914
1915 if (level == SOL_SOCKET)
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001916 err =
1917 sock_setsockopt(sock, level, optname, optval,
1918 optlen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001919 else
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001920 err =
1921 sock->ops->setsockopt(sock, level, optname, optval,
1922 optlen);
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001923out_put:
1924 fput_light(sock->file, fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001925 }
1926 return err;
1927}
1928
1929/*
1930 * Get a socket option. Because we don't know the option lengths we have
1931 * to pass a user mode parameter for the protocols to sort out.
1932 */
1933
Heiko Carstens20f37032009-01-14 14:14:23 +01001934SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname,
1935 char __user *, optval, int __user *, optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001936{
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001937 int err, fput_needed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001938 struct socket *sock;
1939
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001940 sock = sockfd_lookup_light(fd, &err, &fput_needed);
1941 if (sock != NULL) {
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001942 err = security_socket_getsockopt(sock, level, optname);
1943 if (err)
1944 goto out_put;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001945
1946 if (level == SOL_SOCKET)
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001947 err =
1948 sock_getsockopt(sock, level, optname, optval,
1949 optlen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001950 else
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001951 err =
1952 sock->ops->getsockopt(sock, level, optname, optval,
1953 optlen);
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001954out_put:
1955 fput_light(sock->file, fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001956 }
1957 return err;
1958}
1959
Linus Torvalds1da177e2005-04-16 15:20:36 -07001960/*
1961 * Shutdown a socket.
1962 */
1963
Heiko Carstens754fe8d2009-01-14 14:14:09 +01001964SYSCALL_DEFINE2(shutdown, int, fd, int, how)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001965{
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001966 int err, fput_needed;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001967 struct socket *sock;
1968
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001969 sock = sockfd_lookup_light(fd, &err, &fput_needed);
1970 if (sock != NULL) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001971 err = security_socket_shutdown(sock, how);
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08001972 if (!err)
1973 err = sock->ops->shutdown(sock, how);
1974 fput_light(sock->file, fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001975 }
1976 return err;
1977}
1978
Stephen Hemminger89bddce2006-09-01 00:19:31 -07001979/* A couple of helpful macros for getting the address of the 32/64 bit
Linus Torvalds1da177e2005-04-16 15:20:36 -07001980 * fields which are the same type (int / unsigned) on our platforms.
1981 */
1982#define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member)
1983#define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen)
1984#define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags)
1985
Tetsuo Handac71d8eb2011-08-04 14:07:40 +00001986struct used_address {
1987 struct sockaddr_storage name;
1988 unsigned int name_len;
1989};
1990
Dan Carpenter1661bf32013-10-03 00:27:20 +03001991static int copy_msghdr_from_user(struct msghdr *kmsg,
1992 struct msghdr __user *umsg)
1993{
1994 if (copy_from_user(kmsg, umsg, sizeof(struct msghdr)))
1995 return -EFAULT;
Matthew Leachdbb490b2014-03-11 11:58:27 +00001996
Ani Sinha6a2a2b32014-09-08 14:49:59 -07001997 if (kmsg->msg_name == NULL)
1998 kmsg->msg_namelen = 0;
1999
Matthew Leachdbb490b2014-03-11 11:58:27 +00002000 if (kmsg->msg_namelen < 0)
2001 return -EINVAL;
2002
Dan Carpenter1661bf32013-10-03 00:27:20 +03002003 if (kmsg->msg_namelen > sizeof(struct sockaddr_storage))
Dan Carpenterdb31c552013-11-27 15:40:21 +03002004 kmsg->msg_namelen = sizeof(struct sockaddr_storage);
Dan Carpenter1661bf32013-10-03 00:27:20 +03002005 return 0;
2006}
2007
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002008static int ___sys_sendmsg(struct socket *sock, struct msghdr __user *msg,
Eric Dumazet95c96172012-04-15 05:58:06 +00002009 struct msghdr *msg_sys, unsigned int flags,
Tetsuo Handac71d8eb2011-08-04 14:07:40 +00002010 struct used_address *used_address)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002011{
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002012 struct compat_msghdr __user *msg_compat =
2013 (struct compat_msghdr __user *)msg;
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07002014 struct sockaddr_storage address;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002015 struct iovec iovstack[UIO_FASTIOV], *iov = iovstack;
Alex Williamsonb9d717a2005-09-26 14:28:02 -07002016 unsigned char ctl[sizeof(struct cmsghdr) + 20]
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002017 __attribute__ ((aligned(sizeof(__kernel_size_t))));
2018 /* 20 is size of ipv6_pktinfo */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002019 unsigned char *ctl_buf = ctl;
Eric Dumazeta74e9102012-04-20 20:04:01 +02002020 int err, ctl_len, total_len;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002021
Linus Torvalds1da177e2005-04-16 15:20:36 -07002022 err = -EFAULT;
2023 if (MSG_CMSG_COMPAT & flags) {
Anton Blanchard228e5482011-05-02 20:21:35 +00002024 if (get_compat_msghdr(msg_sys, msg_compat))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002025 return -EFAULT;
Dan Carpenter1661bf32013-10-03 00:27:20 +03002026 } else {
2027 err = copy_msghdr_from_user(msg_sys, msg);
2028 if (err)
2029 return err;
2030 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002031
Anton Blanchard228e5482011-05-02 20:21:35 +00002032 if (msg_sys->msg_iovlen > UIO_FASTIOV) {
Eric Dumazeta74e9102012-04-20 20:04:01 +02002033 err = -EMSGSIZE;
2034 if (msg_sys->msg_iovlen > UIO_MAXIOV)
2035 goto out;
2036 err = -ENOMEM;
2037 iov = kmalloc(msg_sys->msg_iovlen * sizeof(struct iovec),
2038 GFP_KERNEL);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002039 if (!iov)
Anton Blanchard228e5482011-05-02 20:21:35 +00002040 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002041 }
2042
2043 /* This will also move the address data into kernel space */
2044 if (MSG_CMSG_COMPAT & flags) {
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00002045 err = verify_compat_iovec(msg_sys, iov, &address, VERIFY_READ);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002046 } else
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00002047 err = verify_iovec(msg_sys, iov, &address, VERIFY_READ);
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002048 if (err < 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002049 goto out_freeiov;
2050 total_len = err;
2051
2052 err = -ENOBUFS;
2053
Anton Blanchard228e5482011-05-02 20:21:35 +00002054 if (msg_sys->msg_controllen > INT_MAX)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002055 goto out_freeiov;
Anton Blanchard228e5482011-05-02 20:21:35 +00002056 ctl_len = msg_sys->msg_controllen;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002057 if ((MSG_CMSG_COMPAT & flags) && ctl_len) {
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002058 err =
Anton Blanchard228e5482011-05-02 20:21:35 +00002059 cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl,
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002060 sizeof(ctl));
Linus Torvalds1da177e2005-04-16 15:20:36 -07002061 if (err)
2062 goto out_freeiov;
Anton Blanchard228e5482011-05-02 20:21:35 +00002063 ctl_buf = msg_sys->msg_control;
2064 ctl_len = msg_sys->msg_controllen;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002065 } else if (ctl_len) {
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002066 if (ctl_len > sizeof(ctl)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002067 ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL);
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002068 if (ctl_buf == NULL)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002069 goto out_freeiov;
2070 }
2071 err = -EFAULT;
2072 /*
Anton Blanchard228e5482011-05-02 20:21:35 +00002073 * Careful! Before this, msg_sys->msg_control contains a user pointer.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002074 * Afterwards, it will be a kernel pointer. Thus the compiler-assisted
2075 * checking falls down on this.
2076 */
Namhyung Kimfb8621b2010-09-07 03:55:00 +00002077 if (copy_from_user(ctl_buf,
Anton Blanchard228e5482011-05-02 20:21:35 +00002078 (void __user __force *)msg_sys->msg_control,
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002079 ctl_len))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002080 goto out_freectl;
Anton Blanchard228e5482011-05-02 20:21:35 +00002081 msg_sys->msg_control = ctl_buf;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002082 }
Anton Blanchard228e5482011-05-02 20:21:35 +00002083 msg_sys->msg_flags = flags;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002084
2085 if (sock->file->f_flags & O_NONBLOCK)
Anton Blanchard228e5482011-05-02 20:21:35 +00002086 msg_sys->msg_flags |= MSG_DONTWAIT;
Tetsuo Handac71d8eb2011-08-04 14:07:40 +00002087 /*
2088 * If this is sendmmsg() and current destination address is same as
2089 * previously succeeded address, omit asking LSM's decision.
2090 * used_address->name_len is initialized to UINT_MAX so that the first
2091 * destination address never matches.
2092 */
Mathieu Desnoyersbc909d92011-08-24 19:45:03 -07002093 if (used_address && msg_sys->msg_name &&
2094 used_address->name_len == msg_sys->msg_namelen &&
2095 !memcmp(&used_address->name, msg_sys->msg_name,
Tetsuo Handac71d8eb2011-08-04 14:07:40 +00002096 used_address->name_len)) {
2097 err = sock_sendmsg_nosec(sock, msg_sys, total_len);
2098 goto out_freectl;
2099 }
2100 err = sock_sendmsg(sock, msg_sys, total_len);
2101 /*
2102 * If this is sendmmsg() and sending to current destination address was
2103 * successful, remember it.
2104 */
2105 if (used_address && err >= 0) {
2106 used_address->name_len = msg_sys->msg_namelen;
Mathieu Desnoyersbc909d92011-08-24 19:45:03 -07002107 if (msg_sys->msg_name)
2108 memcpy(&used_address->name, msg_sys->msg_name,
2109 used_address->name_len);
Tetsuo Handac71d8eb2011-08-04 14:07:40 +00002110 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002111
2112out_freectl:
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002113 if (ctl_buf != ctl)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002114 sock_kfree_s(sock->sk, ctl_buf, ctl_len);
2115out_freeiov:
2116 if (iov != iovstack)
Eric Dumazeta74e9102012-04-20 20:04:01 +02002117 kfree(iov);
Anton Blanchard228e5482011-05-02 20:21:35 +00002118out:
2119 return err;
2120}
2121
2122/*
2123 * BSD sendmsg interface
2124 */
2125
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002126long __sys_sendmsg(int fd, struct msghdr __user *msg, unsigned flags)
Anton Blanchard228e5482011-05-02 20:21:35 +00002127{
2128 int fput_needed, err;
2129 struct msghdr msg_sys;
Andy Lutomirski1be374a2013-05-22 14:07:44 -07002130 struct socket *sock;
Anton Blanchard228e5482011-05-02 20:21:35 +00002131
Andy Lutomirski1be374a2013-05-22 14:07:44 -07002132 sock = sockfd_lookup_light(fd, &err, &fput_needed);
Anton Blanchard228e5482011-05-02 20:21:35 +00002133 if (!sock)
2134 goto out;
2135
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002136 err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL);
Anton Blanchard228e5482011-05-02 20:21:35 +00002137
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08002138 fput_light(sock->file, fput_needed);
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002139out:
Linus Torvalds1da177e2005-04-16 15:20:36 -07002140 return err;
2141}
2142
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002143SYSCALL_DEFINE3(sendmsg, int, fd, struct msghdr __user *, msg, unsigned int, flags)
2144{
2145 if (flags & MSG_CMSG_COMPAT)
2146 return -EINVAL;
2147 return __sys_sendmsg(fd, msg, flags);
2148}
2149
Anton Blanchard228e5482011-05-02 20:21:35 +00002150/*
2151 * Linux sendmmsg interface
2152 */
2153
2154int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen,
2155 unsigned int flags)
2156{
2157 int fput_needed, err, datagrams;
2158 struct socket *sock;
2159 struct mmsghdr __user *entry;
2160 struct compat_mmsghdr __user *compat_entry;
2161 struct msghdr msg_sys;
Tetsuo Handac71d8eb2011-08-04 14:07:40 +00002162 struct used_address used_address;
Anton Blanchard228e5482011-05-02 20:21:35 +00002163
Anton Blanchard98382f42011-08-04 14:07:39 +00002164 if (vlen > UIO_MAXIOV)
2165 vlen = UIO_MAXIOV;
Anton Blanchard228e5482011-05-02 20:21:35 +00002166
2167 datagrams = 0;
2168
2169 sock = sockfd_lookup_light(fd, &err, &fput_needed);
2170 if (!sock)
2171 return err;
2172
Tetsuo Handac71d8eb2011-08-04 14:07:40 +00002173 used_address.name_len = UINT_MAX;
Anton Blanchard228e5482011-05-02 20:21:35 +00002174 entry = mmsg;
2175 compat_entry = (struct compat_mmsghdr __user *)mmsg;
Anton Blanchard728ffb82011-08-04 14:07:38 +00002176 err = 0;
Anton Blanchard228e5482011-05-02 20:21:35 +00002177
2178 while (datagrams < vlen) {
Anton Blanchard228e5482011-05-02 20:21:35 +00002179 if (MSG_CMSG_COMPAT & flags) {
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002180 err = ___sys_sendmsg(sock, (struct msghdr __user *)compat_entry,
2181 &msg_sys, flags, &used_address);
Anton Blanchard228e5482011-05-02 20:21:35 +00002182 if (err < 0)
2183 break;
2184 err = __put_user(err, &compat_entry->msg_len);
2185 ++compat_entry;
2186 } else {
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002187 err = ___sys_sendmsg(sock,
2188 (struct msghdr __user *)entry,
2189 &msg_sys, flags, &used_address);
Anton Blanchard228e5482011-05-02 20:21:35 +00002190 if (err < 0)
2191 break;
2192 err = put_user(err, &entry->msg_len);
2193 ++entry;
2194 }
2195
2196 if (err)
2197 break;
2198 ++datagrams;
2199 }
2200
Anton Blanchard228e5482011-05-02 20:21:35 +00002201 fput_light(sock->file, fput_needed);
2202
Anton Blanchard728ffb82011-08-04 14:07:38 +00002203 /* We only return an error if no datagrams were able to be sent */
2204 if (datagrams != 0)
Anton Blanchard228e5482011-05-02 20:21:35 +00002205 return datagrams;
2206
Anton Blanchard228e5482011-05-02 20:21:35 +00002207 return err;
2208}
2209
2210SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg,
2211 unsigned int, vlen, unsigned int, flags)
2212{
Andy Lutomirski1be374a2013-05-22 14:07:44 -07002213 if (flags & MSG_CMSG_COMPAT)
2214 return -EINVAL;
Anton Blanchard228e5482011-05-02 20:21:35 +00002215 return __sys_sendmmsg(fd, mmsg, vlen, flags);
2216}
2217
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002218static int ___sys_recvmsg(struct socket *sock, struct msghdr __user *msg,
Eric Dumazet95c96172012-04-15 05:58:06 +00002219 struct msghdr *msg_sys, unsigned int flags, int nosec)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002220{
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002221 struct compat_msghdr __user *msg_compat =
2222 (struct compat_msghdr __user *)msg;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002223 struct iovec iovstack[UIO_FASTIOV];
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002224 struct iovec *iov = iovstack;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002225 unsigned long cmsg_ptr;
Eric Dumazeta74e9102012-04-20 20:04:01 +02002226 int err, total_len, len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002227
2228 /* kernel mode address */
YOSHIFUJI Hideaki230b1832008-07-19 22:35:47 -07002229 struct sockaddr_storage addr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002230
2231 /* user mode address pointers */
2232 struct sockaddr __user *uaddr;
2233 int __user *uaddr_len;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002234
Linus Torvalds1da177e2005-04-16 15:20:36 -07002235 if (MSG_CMSG_COMPAT & flags) {
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002236 if (get_compat_msghdr(msg_sys, msg_compat))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002237 return -EFAULT;
Dan Carpenter1661bf32013-10-03 00:27:20 +03002238 } else {
2239 err = copy_msghdr_from_user(msg_sys, msg);
2240 if (err)
2241 return err;
2242 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002243
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002244 if (msg_sys->msg_iovlen > UIO_FASTIOV) {
Eric Dumazeta74e9102012-04-20 20:04:01 +02002245 err = -EMSGSIZE;
2246 if (msg_sys->msg_iovlen > UIO_MAXIOV)
2247 goto out;
2248 err = -ENOMEM;
2249 iov = kmalloc(msg_sys->msg_iovlen * sizeof(struct iovec),
2250 GFP_KERNEL);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002251 if (!iov)
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002252 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002253 }
2254
Hannes Frederic Sowaf3d33422013-11-21 03:14:22 +01002255 /* Save the user-mode address (verify_iovec will change the
2256 * kernel msghdr to use the kernel address space)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002257 */
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002258 uaddr = (__force void __user *)msg_sys->msg_name;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002259 uaddr_len = COMPAT_NAMELEN(msg);
Hannes Frederic Sowaf3d33422013-11-21 03:14:22 +01002260 if (MSG_CMSG_COMPAT & flags)
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00002261 err = verify_compat_iovec(msg_sys, iov, &addr, VERIFY_WRITE);
Hannes Frederic Sowaf3d33422013-11-21 03:14:22 +01002262 else
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00002263 err = verify_iovec(msg_sys, iov, &addr, VERIFY_WRITE);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002264 if (err < 0)
2265 goto out_freeiov;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002266 total_len = err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002267
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002268 cmsg_ptr = (unsigned long)msg_sys->msg_control;
2269 msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT);
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002270
Hannes Frederic Sowaf3d33422013-11-21 03:14:22 +01002271 /* We assume all kernel code knows the size of sockaddr_storage */
2272 msg_sys->msg_namelen = 0;
2273
Linus Torvalds1da177e2005-04-16 15:20:36 -07002274 if (sock->file->f_flags & O_NONBLOCK)
2275 flags |= MSG_DONTWAIT;
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002276 err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys,
2277 total_len, flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002278 if (err < 0)
2279 goto out_freeiov;
2280 len = err;
2281
2282 if (uaddr != NULL) {
Maciej Żenczykowski43db3622012-03-11 12:51:50 +00002283 err = move_addr_to_user(&addr,
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002284 msg_sys->msg_namelen, uaddr,
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002285 uaddr_len);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002286 if (err < 0)
2287 goto out_freeiov;
2288 }
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002289 err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT),
David S. Miller37f7f422005-09-16 16:51:01 -07002290 COMPAT_FLAGS(msg));
Linus Torvalds1da177e2005-04-16 15:20:36 -07002291 if (err)
2292 goto out_freeiov;
2293 if (MSG_CMSG_COMPAT & flags)
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002294 err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002295 &msg_compat->msg_controllen);
2296 else
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002297 err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002298 &msg->msg_controllen);
2299 if (err)
2300 goto out_freeiov;
2301 err = len;
2302
2303out_freeiov:
2304 if (iov != iovstack)
Eric Dumazeta74e9102012-04-20 20:04:01 +02002305 kfree(iov);
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002306out:
2307 return err;
2308}
2309
2310/*
2311 * BSD recvmsg interface
2312 */
2313
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002314long __sys_recvmsg(int fd, struct msghdr __user *msg, unsigned flags)
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002315{
2316 int fput_needed, err;
2317 struct msghdr msg_sys;
Andy Lutomirski1be374a2013-05-22 14:07:44 -07002318 struct socket *sock;
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002319
Andy Lutomirski1be374a2013-05-22 14:07:44 -07002320 sock = sockfd_lookup_light(fd, &err, &fput_needed);
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002321 if (!sock)
2322 goto out;
2323
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002324 err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0);
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002325
Benjamin LaHaise6cb153c2006-03-20 22:27:12 -08002326 fput_light(sock->file, fput_needed);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002327out:
2328 return err;
2329}
2330
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002331SYSCALL_DEFINE3(recvmsg, int, fd, struct msghdr __user *, msg,
2332 unsigned int, flags)
2333{
2334 if (flags & MSG_CMSG_COMPAT)
2335 return -EINVAL;
2336 return __sys_recvmsg(fd, msg, flags);
2337}
2338
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002339/*
2340 * Linux recvmmsg interface
2341 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002342
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002343int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen,
2344 unsigned int flags, struct timespec *timeout)
2345{
2346 int fput_needed, err, datagrams;
2347 struct socket *sock;
2348 struct mmsghdr __user *entry;
Jean-Mickael Guerind7256d02009-12-01 08:47:26 +00002349 struct compat_mmsghdr __user *compat_entry;
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002350 struct msghdr msg_sys;
2351 struct timespec end_time;
2352
2353 if (timeout &&
2354 poll_select_set_timeout(&end_time, timeout->tv_sec,
2355 timeout->tv_nsec))
2356 return -EINVAL;
2357
2358 datagrams = 0;
2359
2360 sock = sockfd_lookup_light(fd, &err, &fput_needed);
2361 if (!sock)
2362 return err;
2363
2364 err = sock_error(sock->sk);
2365 if (err)
2366 goto out_put;
2367
2368 entry = mmsg;
Jean-Mickael Guerind7256d02009-12-01 08:47:26 +00002369 compat_entry = (struct compat_mmsghdr __user *)mmsg;
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002370
2371 while (datagrams < vlen) {
2372 /*
2373 * No need to ask LSM for more than the first datagram.
2374 */
Jean-Mickael Guerind7256d02009-12-01 08:47:26 +00002375 if (MSG_CMSG_COMPAT & flags) {
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002376 err = ___sys_recvmsg(sock, (struct msghdr __user *)compat_entry,
2377 &msg_sys, flags & ~MSG_WAITFORONE,
2378 datagrams);
Jean-Mickael Guerind7256d02009-12-01 08:47:26 +00002379 if (err < 0)
2380 break;
2381 err = __put_user(err, &compat_entry->msg_len);
2382 ++compat_entry;
2383 } else {
Andy Lutomirskia7526eb2013-06-05 19:38:26 +00002384 err = ___sys_recvmsg(sock,
2385 (struct msghdr __user *)entry,
2386 &msg_sys, flags & ~MSG_WAITFORONE,
2387 datagrams);
Jean-Mickael Guerind7256d02009-12-01 08:47:26 +00002388 if (err < 0)
2389 break;
2390 err = put_user(err, &entry->msg_len);
2391 ++entry;
2392 }
2393
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002394 if (err)
2395 break;
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002396 ++datagrams;
2397
Brandon L Black71c5c1592010-03-26 16:18:03 +00002398 /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */
2399 if (flags & MSG_WAITFORONE)
2400 flags |= MSG_DONTWAIT;
2401
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002402 if (timeout) {
2403 ktime_get_ts(timeout);
2404 *timeout = timespec_sub(end_time, *timeout);
2405 if (timeout->tv_sec < 0) {
2406 timeout->tv_sec = timeout->tv_nsec = 0;
2407 break;
2408 }
2409
2410 /* Timeout, return less than vlen datagrams */
2411 if (timeout->tv_nsec == 0 && timeout->tv_sec == 0)
2412 break;
2413 }
2414
2415 /* Out of band data, return right away */
2416 if (msg_sys.msg_flags & MSG_OOB)
2417 break;
2418 }
2419
2420out_put:
2421 fput_light(sock->file, fput_needed);
2422
2423 if (err == 0)
2424 return datagrams;
2425
2426 if (datagrams != 0) {
2427 /*
2428 * We may return less entries than requested (vlen) if the
2429 * sock is non block and there aren't enough datagrams...
2430 */
2431 if (err != -EAGAIN) {
2432 /*
2433 * ... or if recvmsg returns an error after we
2434 * received some datagrams, where we record the
2435 * error to return on the next call or if the
2436 * app asks about it using getsockopt(SO_ERROR).
2437 */
2438 sock->sk->sk_err = -err;
2439 }
2440
2441 return datagrams;
2442 }
2443
2444 return err;
2445}
2446
2447SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg,
2448 unsigned int, vlen, unsigned int, flags,
2449 struct timespec __user *, timeout)
2450{
2451 int datagrams;
2452 struct timespec timeout_sys;
2453
Andy Lutomirski1be374a2013-05-22 14:07:44 -07002454 if (flags & MSG_CMSG_COMPAT)
2455 return -EINVAL;
2456
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002457 if (!timeout)
2458 return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL);
2459
2460 if (copy_from_user(&timeout_sys, timeout, sizeof(timeout_sys)))
2461 return -EFAULT;
2462
2463 datagrams = __sys_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys);
2464
2465 if (datagrams > 0 &&
2466 copy_to_user(timeout, &timeout_sys, sizeof(timeout_sys)))
2467 datagrams = -EFAULT;
2468
2469 return datagrams;
2470}
2471
2472#ifdef __ARCH_WANT_SYS_SOCKETCALL
Linus Torvalds1da177e2005-04-16 15:20:36 -07002473/* Argument list sizes for sys_socketcall */
2474#define AL(x) ((x) * sizeof(unsigned long))
Anton Blanchard228e5482011-05-02 20:21:35 +00002475static const unsigned char nargs[21] = {
Eric Dumazetc6d409c2010-06-03 20:03:40 -07002476 AL(0), AL(3), AL(3), AL(3), AL(2), AL(3),
2477 AL(3), AL(3), AL(4), AL(4), AL(4), AL(6),
2478 AL(6), AL(2), AL(5), AL(5), AL(3), AL(3),
Anton Blanchard228e5482011-05-02 20:21:35 +00002479 AL(4), AL(5), AL(4)
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002480};
2481
Linus Torvalds1da177e2005-04-16 15:20:36 -07002482#undef AL
2483
2484/*
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002485 * System call vectors.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002486 *
2487 * Argument checking cleaned up. Saved 20% in size.
2488 * This function doesn't need to set the kernel lock because
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002489 * it is set by the callees.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002490 */
2491
Heiko Carstens3e0fa652009-01-14 14:14:24 +01002492SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002493{
Chen Gang2950fa92013-04-07 16:55:23 +08002494 unsigned long a[AUDITSC_ARGS];
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002495 unsigned long a0, a1;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002496 int err;
Arjan van de Ven47379052009-09-28 12:57:44 -07002497 unsigned int len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002498
Anton Blanchard228e5482011-05-02 20:21:35 +00002499 if (call < 1 || call > SYS_SENDMMSG)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002500 return -EINVAL;
2501
Arjan van de Ven47379052009-09-28 12:57:44 -07002502 len = nargs[call];
2503 if (len > sizeof(a))
2504 return -EINVAL;
2505
Linus Torvalds1da177e2005-04-16 15:20:36 -07002506 /* copy_from_user should be SMP safe. */
Arjan van de Ven47379052009-09-28 12:57:44 -07002507 if (copy_from_user(a, args, len))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002508 return -EFAULT;
David Woodhouse3ec3b2f2005-05-17 12:08:48 +01002509
Chen Gang2950fa92013-04-07 16:55:23 +08002510 err = audit_socketcall(nargs[call] / sizeof(unsigned long), a);
2511 if (err)
2512 return err;
David Woodhouse3ec3b2f2005-05-17 12:08:48 +01002513
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002514 a0 = a[0];
2515 a1 = a[1];
2516
2517 switch (call) {
2518 case SYS_SOCKET:
2519 err = sys_socket(a0, a1, a[2]);
2520 break;
2521 case SYS_BIND:
2522 err = sys_bind(a0, (struct sockaddr __user *)a1, a[2]);
2523 break;
2524 case SYS_CONNECT:
2525 err = sys_connect(a0, (struct sockaddr __user *)a1, a[2]);
2526 break;
2527 case SYS_LISTEN:
2528 err = sys_listen(a0, a1);
2529 break;
2530 case SYS_ACCEPT:
Ulrich Drepperde11def2008-11-19 15:36:14 -08002531 err = sys_accept4(a0, (struct sockaddr __user *)a1,
2532 (int __user *)a[2], 0);
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002533 break;
2534 case SYS_GETSOCKNAME:
2535 err =
2536 sys_getsockname(a0, (struct sockaddr __user *)a1,
2537 (int __user *)a[2]);
2538 break;
2539 case SYS_GETPEERNAME:
2540 err =
2541 sys_getpeername(a0, (struct sockaddr __user *)a1,
2542 (int __user *)a[2]);
2543 break;
2544 case SYS_SOCKETPAIR:
2545 err = sys_socketpair(a0, a1, a[2], (int __user *)a[3]);
2546 break;
2547 case SYS_SEND:
2548 err = sys_send(a0, (void __user *)a1, a[2], a[3]);
2549 break;
2550 case SYS_SENDTO:
2551 err = sys_sendto(a0, (void __user *)a1, a[2], a[3],
2552 (struct sockaddr __user *)a[4], a[5]);
2553 break;
2554 case SYS_RECV:
2555 err = sys_recv(a0, (void __user *)a1, a[2], a[3]);
2556 break;
2557 case SYS_RECVFROM:
2558 err = sys_recvfrom(a0, (void __user *)a1, a[2], a[3],
2559 (struct sockaddr __user *)a[4],
2560 (int __user *)a[5]);
2561 break;
2562 case SYS_SHUTDOWN:
2563 err = sys_shutdown(a0, a1);
2564 break;
2565 case SYS_SETSOCKOPT:
2566 err = sys_setsockopt(a0, a1, a[2], (char __user *)a[3], a[4]);
2567 break;
2568 case SYS_GETSOCKOPT:
2569 err =
2570 sys_getsockopt(a0, a1, a[2], (char __user *)a[3],
2571 (int __user *)a[4]);
2572 break;
2573 case SYS_SENDMSG:
2574 err = sys_sendmsg(a0, (struct msghdr __user *)a1, a[2]);
2575 break;
Anton Blanchard228e5482011-05-02 20:21:35 +00002576 case SYS_SENDMMSG:
2577 err = sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3]);
2578 break;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002579 case SYS_RECVMSG:
2580 err = sys_recvmsg(a0, (struct msghdr __user *)a1, a[2]);
2581 break;
Arnaldo Carvalho de Meloa2e27252009-10-12 23:40:10 -07002582 case SYS_RECVMMSG:
2583 err = sys_recvmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3],
2584 (struct timespec __user *)a[4]);
2585 break;
Ulrich Drepperde11def2008-11-19 15:36:14 -08002586 case SYS_ACCEPT4:
2587 err = sys_accept4(a0, (struct sockaddr __user *)a1,
2588 (int __user *)a[2], a[3]);
Ulrich Drepperaaca0bd2008-07-23 21:29:20 -07002589 break;
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002590 default:
2591 err = -EINVAL;
2592 break;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002593 }
2594 return err;
2595}
2596
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002597#endif /* __ARCH_WANT_SYS_SOCKETCALL */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002598
Stephen Hemminger55737fd2006-09-01 00:23:39 -07002599/**
2600 * sock_register - add a socket protocol handler
2601 * @ops: description of protocol
2602 *
Linus Torvalds1da177e2005-04-16 15:20:36 -07002603 * This function is called by a protocol handler that wants to
2604 * advertise its address family, and have it linked into the
Masanari Iidae793c0f2014-09-04 23:44:36 +09002605 * socket interface. The value ops->family corresponds to the
Stephen Hemminger55737fd2006-09-01 00:23:39 -07002606 * socket system call protocol family.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002607 */
Stephen Hemmingerf0fd27d2006-08-09 21:03:17 -07002608int sock_register(const struct net_proto_family *ops)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002609{
2610 int err;
2611
2612 if (ops->family >= NPROTO) {
Yang Yingliang3410f222014-02-12 17:09:55 +08002613 pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002614 return -ENOBUFS;
2615 }
Stephen Hemminger55737fd2006-09-01 00:23:39 -07002616
2617 spin_lock(&net_family_lock);
Eric Dumazet190683a2010-11-10 10:50:44 +00002618 if (rcu_dereference_protected(net_families[ops->family],
2619 lockdep_is_held(&net_family_lock)))
Stephen Hemminger55737fd2006-09-01 00:23:39 -07002620 err = -EEXIST;
2621 else {
Eric Dumazetcf778b02012-01-12 04:41:32 +00002622 rcu_assign_pointer(net_families[ops->family], ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002623 err = 0;
2624 }
Stephen Hemminger55737fd2006-09-01 00:23:39 -07002625 spin_unlock(&net_family_lock);
2626
Yang Yingliang3410f222014-02-12 17:09:55 +08002627 pr_info("NET: Registered protocol family %d\n", ops->family);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002628 return err;
2629}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07002630EXPORT_SYMBOL(sock_register);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002631
Stephen Hemminger55737fd2006-09-01 00:23:39 -07002632/**
2633 * sock_unregister - remove a protocol handler
2634 * @family: protocol family to remove
2635 *
Linus Torvalds1da177e2005-04-16 15:20:36 -07002636 * This function is called by a protocol handler that wants to
2637 * remove its address family, and have it unlinked from the
Stephen Hemminger55737fd2006-09-01 00:23:39 -07002638 * new socket creation.
2639 *
2640 * If protocol handler is a module, then it can use module reference
2641 * counts to protect against new references. If protocol handler is not
2642 * a module then it needs to provide its own protection in
2643 * the ops->create routine.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002644 */
Stephen Hemmingerf0fd27d2006-08-09 21:03:17 -07002645void sock_unregister(int family)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002646{
Stephen Hemmingerf0fd27d2006-08-09 21:03:17 -07002647 BUG_ON(family < 0 || family >= NPROTO);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002648
Stephen Hemminger55737fd2006-09-01 00:23:39 -07002649 spin_lock(&net_family_lock);
Stephen Hemmingera9b3cd72011-08-01 16:19:00 +00002650 RCU_INIT_POINTER(net_families[family], NULL);
Stephen Hemminger55737fd2006-09-01 00:23:39 -07002651 spin_unlock(&net_family_lock);
2652
2653 synchronize_rcu();
2654
Yang Yingliang3410f222014-02-12 17:09:55 +08002655 pr_info("NET: Unregistered protocol family %d\n", family);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002656}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07002657EXPORT_SYMBOL(sock_unregister);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002658
Andi Kleen77d76ea2005-12-22 12:43:42 -08002659static int __init sock_init(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002660{
Nick Pigginb3e19d92011-01-07 17:50:11 +11002661 int err;
Eric W. Biederman2ca794e2012-04-19 13:20:32 +00002662 /*
2663 * Initialize the network sysctl infrastructure.
2664 */
2665 err = net_sysctl_init();
2666 if (err)
2667 goto out;
Nick Pigginb3e19d92011-01-07 17:50:11 +11002668
Linus Torvalds1da177e2005-04-16 15:20:36 -07002669 /*
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002670 * Initialize skbuff SLAB cache
Linus Torvalds1da177e2005-04-16 15:20:36 -07002671 */
2672 skb_init();
Linus Torvalds1da177e2005-04-16 15:20:36 -07002673
2674 /*
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002675 * Initialize the protocols module.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002676 */
2677
2678 init_inodecache();
Nick Pigginb3e19d92011-01-07 17:50:11 +11002679
2680 err = register_filesystem(&sock_fs_type);
2681 if (err)
2682 goto out_fs;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002683 sock_mnt = kern_mount(&sock_fs_type);
Nick Pigginb3e19d92011-01-07 17:50:11 +11002684 if (IS_ERR(sock_mnt)) {
2685 err = PTR_ERR(sock_mnt);
2686 goto out_mount;
2687 }
Andi Kleen77d76ea2005-12-22 12:43:42 -08002688
2689 /* The real protocol initialization is performed in later initcalls.
Linus Torvalds1da177e2005-04-16 15:20:36 -07002690 */
2691
2692#ifdef CONFIG_NETFILTER
Pablo Neira Ayuso6d11cfd2013-05-22 22:42:36 +00002693 err = netfilter_init();
2694 if (err)
2695 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002696#endif
David S. Millercbeb3212005-12-22 12:58:55 -08002697
Daniel Borkmann408eccc2014-04-01 16:20:23 +02002698 ptp_classifier_init();
Richard Cochranc1f19b52010-07-17 08:49:36 +00002699
Nick Pigginb3e19d92011-01-07 17:50:11 +11002700out:
2701 return err;
2702
2703out_mount:
2704 unregister_filesystem(&sock_fs_type);
2705out_fs:
2706 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002707}
2708
Andi Kleen77d76ea2005-12-22 12:43:42 -08002709core_initcall(sock_init); /* early initcall */
2710
Linus Torvalds1da177e2005-04-16 15:20:36 -07002711#ifdef CONFIG_PROC_FS
2712void socket_seq_show(struct seq_file *seq)
2713{
2714 int cpu;
2715 int counter = 0;
2716
KAMEZAWA Hiroyuki6f912042006-04-10 22:52:50 -07002717 for_each_possible_cpu(cpu)
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002718 counter += per_cpu(sockets_in_use, cpu);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002719
2720 /* It can be negative, by the way. 8) */
2721 if (counter < 0)
2722 counter = 0;
2723
2724 seq_printf(seq, "sockets: used %d\n", counter);
2725}
Stephen Hemminger89bddce2006-09-01 00:19:31 -07002726#endif /* CONFIG_PROC_FS */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002727
Shaun Pereira89bbfc92006-03-21 23:58:08 -08002728#ifdef CONFIG_COMPAT
Arnd Bergmann6b960182009-11-06 23:10:54 -08002729static int do_siocgstamp(struct net *net, struct socket *sock,
H. Peter Anvin644595f2012-02-19 17:51:59 -08002730 unsigned int cmd, void __user *up)
Arnd Bergmann7a229382009-11-06 23:00:29 -08002731{
Arnd Bergmann7a229382009-11-06 23:00:29 -08002732 mm_segment_t old_fs = get_fs();
2733 struct timeval ktv;
2734 int err;
2735
2736 set_fs(KERNEL_DS);
Arnd Bergmann6b960182009-11-06 23:10:54 -08002737 err = sock_do_ioctl(net, sock, cmd, (unsigned long)&ktv);
Arnd Bergmann7a229382009-11-06 23:00:29 -08002738 set_fs(old_fs);
H. Peter Anvin644595f2012-02-19 17:51:59 -08002739 if (!err)
Mikulas Patockaed6fe9d2012-09-01 12:34:07 -04002740 err = compat_put_timeval(&ktv, up);
H. Peter Anvin644595f2012-02-19 17:51:59 -08002741
Arnd Bergmann7a229382009-11-06 23:00:29 -08002742 return err;
2743}
2744
Arnd Bergmann6b960182009-11-06 23:10:54 -08002745static int do_siocgstampns(struct net *net, struct socket *sock,
H. Peter Anvin644595f2012-02-19 17:51:59 -08002746 unsigned int cmd, void __user *up)
Arnd Bergmann7a229382009-11-06 23:00:29 -08002747{
Arnd Bergmann7a229382009-11-06 23:00:29 -08002748 mm_segment_t old_fs = get_fs();
2749 struct timespec kts;
2750 int err;
2751
2752 set_fs(KERNEL_DS);
Arnd Bergmann6b960182009-11-06 23:10:54 -08002753 err = sock_do_ioctl(net, sock, cmd, (unsigned long)&kts);
Arnd Bergmann7a229382009-11-06 23:00:29 -08002754 set_fs(old_fs);
H. Peter Anvin644595f2012-02-19 17:51:59 -08002755 if (!err)
Mikulas Patockaed6fe9d2012-09-01 12:34:07 -04002756 err = compat_put_timespec(&kts, up);
H. Peter Anvin644595f2012-02-19 17:51:59 -08002757
Arnd Bergmann7a229382009-11-06 23:00:29 -08002758 return err;
2759}
2760
Arnd Bergmann6b960182009-11-06 23:10:54 -08002761static int dev_ifname32(struct net *net, struct compat_ifreq __user *uifr32)
Arnd Bergmann7a229382009-11-06 23:00:29 -08002762{
2763 struct ifreq __user *uifr;
2764 int err;
2765
2766 uifr = compat_alloc_user_space(sizeof(struct ifreq));
Arnd Bergmann6b960182009-11-06 23:10:54 -08002767 if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq)))
Arnd Bergmann7a229382009-11-06 23:00:29 -08002768 return -EFAULT;
2769
Arnd Bergmann6b960182009-11-06 23:10:54 -08002770 err = dev_ioctl(net, SIOCGIFNAME, uifr);
Arnd Bergmann7a229382009-11-06 23:00:29 -08002771 if (err)
2772 return err;
2773
Arnd Bergmann6b960182009-11-06 23:10:54 -08002774 if (copy_in_user(uifr32, uifr, sizeof(struct compat_ifreq)))
Arnd Bergmann7a229382009-11-06 23:00:29 -08002775 return -EFAULT;
2776
2777 return 0;
2778}
2779
Arnd Bergmann6b960182009-11-06 23:10:54 -08002780static int dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32)
Arnd Bergmann7a229382009-11-06 23:00:29 -08002781{
Arnd Bergmann6b960182009-11-06 23:10:54 -08002782 struct compat_ifconf ifc32;
Arnd Bergmann7a229382009-11-06 23:00:29 -08002783 struct ifconf ifc;
2784 struct ifconf __user *uifc;
Arnd Bergmann6b960182009-11-06 23:10:54 -08002785 struct compat_ifreq __user *ifr32;
Arnd Bergmann7a229382009-11-06 23:00:29 -08002786 struct ifreq __user *ifr;
2787 unsigned int i, j;
2788 int err;
2789
Arnd Bergmann6b960182009-11-06 23:10:54 -08002790 if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf)))
Arnd Bergmann7a229382009-11-06 23:00:29 -08002791 return -EFAULT;
2792
Mathias Krause43da5f22012-08-15 11:31:57 +00002793 memset(&ifc, 0, sizeof(ifc));
Arnd Bergmann7a229382009-11-06 23:00:29 -08002794 if (ifc32.ifcbuf == 0) {
2795 ifc32.ifc_len = 0;
2796 ifc.ifc_len = 0;
2797 ifc.ifc_req = NULL;
2798 uifc = compat_alloc_user_space(sizeof(struct ifconf));
2799 } else {
Eric Dumazetc6d409c2010-06-03 20:03:40 -07002800 size_t len = ((ifc32.ifc_len / sizeof(struct compat_ifreq)) + 1) *
2801 sizeof(struct ifreq);
Arnd Bergmann7a229382009-11-06 23:00:29 -08002802 uifc = compat_alloc_user_space(sizeof(struct ifconf) + len);
2803 ifc.ifc_len = len;
2804 ifr = ifc.ifc_req = (void __user *)(uifc + 1);
2805 ifr32 = compat_ptr(ifc32.ifcbuf);
Eric Dumazetc6d409c2010-06-03 20:03:40 -07002806 for (i = 0; i < ifc32.ifc_len; i += sizeof(struct compat_ifreq)) {
Arnd Bergmann6b960182009-11-06 23:10:54 -08002807 if (copy_in_user(ifr, ifr32, sizeof(struct compat_ifreq)))
Arnd Bergmann7a229382009-11-06 23:00:29 -08002808 return -EFAULT;
2809 ifr++;
2810 ifr32++;
2811 }
2812 }
2813 if (copy_to_user(uifc, &ifc, sizeof(struct ifconf)))
2814 return -EFAULT;
2815
Arnd Bergmann6b960182009-11-06 23:10:54 -08002816 err = dev_ioctl(net, SIOCGIFCONF, uifc);
Arnd Bergmann7a229382009-11-06 23:00:29 -08002817 if (err)
2818 return err;
2819
2820 if (copy_from_user(&ifc, uifc, sizeof(struct ifconf)))
2821 return -EFAULT;
2822
2823 ifr = ifc.ifc_req;
2824 ifr32 = compat_ptr(ifc32.ifcbuf);
2825 for (i = 0, j = 0;
Eric Dumazetc6d409c2010-06-03 20:03:40 -07002826 i + sizeof(struct compat_ifreq) <= ifc32.ifc_len && j < ifc.ifc_len;
2827 i += sizeof(struct compat_ifreq), j += sizeof(struct ifreq)) {
2828 if (copy_in_user(ifr32, ifr, sizeof(struct compat_ifreq)))
Arnd Bergmann7a229382009-11-06 23:00:29 -08002829 return -EFAULT;
2830 ifr32++;
2831 ifr++;
2832 }
2833
2834 if (ifc32.ifcbuf == 0) {
2835 /* Translate from 64-bit structure multiple to
2836 * a 32-bit one.
2837 */
2838 i = ifc.ifc_len;
Arnd Bergmann6b960182009-11-06 23:10:54 -08002839 i = ((i / sizeof(struct ifreq)) * sizeof(struct compat_ifreq));
Arnd Bergmann7a229382009-11-06 23:00:29 -08002840 ifc32.ifc_len = i;
2841 } else {
2842 ifc32.ifc_len = i;
2843 }
Arnd Bergmann6b960182009-11-06 23:10:54 -08002844 if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf)))
Arnd Bergmann7a229382009-11-06 23:00:29 -08002845 return -EFAULT;
2846
2847 return 0;
2848}
2849
Arnd Bergmann6b960182009-11-06 23:10:54 -08002850static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32)
Arnd Bergmann7a229382009-11-06 23:00:29 -08002851{
Ben Hutchings3a7da392011-03-17 07:34:32 +00002852 struct compat_ethtool_rxnfc __user *compat_rxnfc;
2853 bool convert_in = false, convert_out = false;
2854 size_t buf_size = ALIGN(sizeof(struct ifreq), 8);
2855 struct ethtool_rxnfc __user *rxnfc;
Arnd Bergmann7a229382009-11-06 23:00:29 -08002856 struct ifreq __user *ifr;
Ben Hutchings3a7da392011-03-17 07:34:32 +00002857 u32 rule_cnt = 0, actual_rule_cnt;
2858 u32 ethcmd;
Arnd Bergmann7a229382009-11-06 23:00:29 -08002859 u32 data;
Ben Hutchings3a7da392011-03-17 07:34:32 +00002860 int ret;
Arnd Bergmann7a229382009-11-06 23:00:29 -08002861
2862 if (get_user(data, &ifr32->ifr_ifru.ifru_data))
2863 return -EFAULT;
2864
Ben Hutchings3a7da392011-03-17 07:34:32 +00002865 compat_rxnfc = compat_ptr(data);
2866
2867 if (get_user(ethcmd, &compat_rxnfc->cmd))
Arnd Bergmann7a229382009-11-06 23:00:29 -08002868 return -EFAULT;
2869
Ben Hutchings3a7da392011-03-17 07:34:32 +00002870 /* Most ethtool structures are defined without padding.
2871 * Unfortunately struct ethtool_rxnfc is an exception.
2872 */
2873 switch (ethcmd) {
2874 default:
2875 break;
2876 case ETHTOOL_GRXCLSRLALL:
2877 /* Buffer size is variable */
2878 if (get_user(rule_cnt, &compat_rxnfc->rule_cnt))
2879 return -EFAULT;
2880 if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32))
2881 return -ENOMEM;
2882 buf_size += rule_cnt * sizeof(u32);
2883 /* fall through */
2884 case ETHTOOL_GRXRINGS:
2885 case ETHTOOL_GRXCLSRLCNT:
2886 case ETHTOOL_GRXCLSRULE:
Ben Hutchings55664f32012-01-03 12:04:51 +00002887 case ETHTOOL_SRXCLSRLINS:
Ben Hutchings3a7da392011-03-17 07:34:32 +00002888 convert_out = true;
2889 /* fall through */
2890 case ETHTOOL_SRXCLSRLDEL:
Ben Hutchings3a7da392011-03-17 07:34:32 +00002891 buf_size += sizeof(struct ethtool_rxnfc);
2892 convert_in = true;
2893 break;
2894 }
2895
2896 ifr = compat_alloc_user_space(buf_size);
Stephen Hemminger954b1242013-02-11 06:22:28 +00002897 rxnfc = (void __user *)ifr + ALIGN(sizeof(struct ifreq), 8);
Ben Hutchings3a7da392011-03-17 07:34:32 +00002898
2899 if (copy_in_user(&ifr->ifr_name, &ifr32->ifr_name, IFNAMSIZ))
2900 return -EFAULT;
2901
2902 if (put_user(convert_in ? rxnfc : compat_ptr(data),
2903 &ifr->ifr_ifru.ifru_data))
2904 return -EFAULT;
2905
2906 if (convert_in) {
Alexander Duyck127fe532011-04-08 18:01:59 +00002907 /* We expect there to be holes between fs.m_ext and
Ben Hutchings3a7da392011-03-17 07:34:32 +00002908 * fs.ring_cookie and at the end of fs, but nowhere else.
2909 */
Alexander Duyck127fe532011-04-08 18:01:59 +00002910 BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) +
2911 sizeof(compat_rxnfc->fs.m_ext) !=
2912 offsetof(struct ethtool_rxnfc, fs.m_ext) +
2913 sizeof(rxnfc->fs.m_ext));
Ben Hutchings3a7da392011-03-17 07:34:32 +00002914 BUILD_BUG_ON(
2915 offsetof(struct compat_ethtool_rxnfc, fs.location) -
2916 offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) !=
2917 offsetof(struct ethtool_rxnfc, fs.location) -
2918 offsetof(struct ethtool_rxnfc, fs.ring_cookie));
2919
2920 if (copy_in_user(rxnfc, compat_rxnfc,
Stephen Hemminger954b1242013-02-11 06:22:28 +00002921 (void __user *)(&rxnfc->fs.m_ext + 1) -
2922 (void __user *)rxnfc) ||
Ben Hutchings3a7da392011-03-17 07:34:32 +00002923 copy_in_user(&rxnfc->fs.ring_cookie,
2924 &compat_rxnfc->fs.ring_cookie,
Stephen Hemminger954b1242013-02-11 06:22:28 +00002925 (void __user *)(&rxnfc->fs.location + 1) -
2926 (void __user *)&rxnfc->fs.ring_cookie) ||
Ben Hutchings3a7da392011-03-17 07:34:32 +00002927 copy_in_user(&rxnfc->rule_cnt, &compat_rxnfc->rule_cnt,
2928 sizeof(rxnfc->rule_cnt)))
2929 return -EFAULT;
2930 }
2931
2932 ret = dev_ioctl(net, SIOCETHTOOL, ifr);
2933 if (ret)
2934 return ret;
2935
2936 if (convert_out) {
2937 if (copy_in_user(compat_rxnfc, rxnfc,
Stephen Hemminger954b1242013-02-11 06:22:28 +00002938 (const void __user *)(&rxnfc->fs.m_ext + 1) -
2939 (const void __user *)rxnfc) ||
Ben Hutchings3a7da392011-03-17 07:34:32 +00002940 copy_in_user(&compat_rxnfc->fs.ring_cookie,
2941 &rxnfc->fs.ring_cookie,
Stephen Hemminger954b1242013-02-11 06:22:28 +00002942 (const void __user *)(&rxnfc->fs.location + 1) -
2943 (const void __user *)&rxnfc->fs.ring_cookie) ||
Ben Hutchings3a7da392011-03-17 07:34:32 +00002944 copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt,
2945 sizeof(rxnfc->rule_cnt)))
2946 return -EFAULT;
2947
2948 if (ethcmd == ETHTOOL_GRXCLSRLALL) {
2949 /* As an optimisation, we only copy the actual
2950 * number of rules that the underlying
2951 * function returned. Since Mallory might
2952 * change the rule count in user memory, we
2953 * check that it is less than the rule count
2954 * originally given (as the user buffer size),
2955 * which has been range-checked.
2956 */
2957 if (get_user(actual_rule_cnt, &rxnfc->rule_cnt))
2958 return -EFAULT;
2959 if (actual_rule_cnt < rule_cnt)
2960 rule_cnt = actual_rule_cnt;
2961 if (copy_in_user(&compat_rxnfc->rule_locs[0],
2962 &rxnfc->rule_locs[0],
2963 rule_cnt * sizeof(u32)))
2964 return -EFAULT;
2965 }
2966 }
2967
2968 return 0;
Arnd Bergmann7a229382009-11-06 23:00:29 -08002969}
2970
Arnd Bergmann7a50a242009-11-08 20:57:03 -08002971static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32)
2972{
2973 void __user *uptr;
2974 compat_uptr_t uptr32;
2975 struct ifreq __user *uifr;
2976
Eric Dumazetc6d409c2010-06-03 20:03:40 -07002977 uifr = compat_alloc_user_space(sizeof(*uifr));
Arnd Bergmann7a50a242009-11-08 20:57:03 -08002978 if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq)))
2979 return -EFAULT;
2980
2981 if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu))
2982 return -EFAULT;
2983
2984 uptr = compat_ptr(uptr32);
2985
2986 if (put_user(uptr, &uifr->ifr_settings.ifs_ifsu.raw_hdlc))
2987 return -EFAULT;
2988
2989 return dev_ioctl(net, SIOCWANDEV, uifr);
2990}
2991
Arnd Bergmann6b960182009-11-06 23:10:54 -08002992static int bond_ioctl(struct net *net, unsigned int cmd,
2993 struct compat_ifreq __user *ifr32)
Arnd Bergmann7a229382009-11-06 23:00:29 -08002994{
2995 struct ifreq kifr;
Arnd Bergmann7a229382009-11-06 23:00:29 -08002996 mm_segment_t old_fs;
2997 int err;
Arnd Bergmann7a229382009-11-06 23:00:29 -08002998
2999 switch (cmd) {
3000 case SIOCBONDENSLAVE:
3001 case SIOCBONDRELEASE:
3002 case SIOCBONDSETHWADDR:
3003 case SIOCBONDCHANGEACTIVE:
Arnd Bergmann6b960182009-11-06 23:10:54 -08003004 if (copy_from_user(&kifr, ifr32, sizeof(struct compat_ifreq)))
Arnd Bergmann7a229382009-11-06 23:00:29 -08003005 return -EFAULT;
3006
3007 old_fs = get_fs();
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003008 set_fs(KERNEL_DS);
stephen hemmingerc3f52ae2011-02-23 09:06:48 +00003009 err = dev_ioctl(net, cmd,
3010 (struct ifreq __user __force *) &kifr);
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003011 set_fs(old_fs);
Arnd Bergmann7a229382009-11-06 23:00:29 -08003012
3013 return err;
Arnd Bergmann7a229382009-11-06 23:00:29 -08003014 default:
Linus Torvalds07d106d2012-01-05 15:40:12 -08003015 return -ENOIOCTLCMD;
Joe Perchesccbd6a52010-05-14 10:58:26 +00003016 }
Arnd Bergmann7a229382009-11-06 23:00:29 -08003017}
3018
Ben Hutchings590d4692013-11-18 17:04:13 +00003019/* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */
3020static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd,
Arnd Bergmann6b960182009-11-06 23:10:54 -08003021 struct compat_ifreq __user *u_ifreq32)
Arnd Bergmann7a229382009-11-06 23:00:29 -08003022{
3023 struct ifreq __user *u_ifreq64;
Arnd Bergmann7a229382009-11-06 23:00:29 -08003024 char tmp_buf[IFNAMSIZ];
3025 void __user *data64;
3026 u32 data32;
3027
3028 if (copy_from_user(&tmp_buf[0], &(u_ifreq32->ifr_ifrn.ifrn_name[0]),
3029 IFNAMSIZ))
3030 return -EFAULT;
Ben Hutchings417c3522013-11-18 17:04:58 +00003031 if (get_user(data32, &u_ifreq32->ifr_ifru.ifru_data))
Arnd Bergmann7a229382009-11-06 23:00:29 -08003032 return -EFAULT;
3033 data64 = compat_ptr(data32);
3034
3035 u_ifreq64 = compat_alloc_user_space(sizeof(*u_ifreq64));
3036
Arnd Bergmann7a229382009-11-06 23:00:29 -08003037 if (copy_to_user(&u_ifreq64->ifr_ifrn.ifrn_name[0], &tmp_buf[0],
3038 IFNAMSIZ))
3039 return -EFAULT;
Ben Hutchings417c3522013-11-18 17:04:58 +00003040 if (put_user(data64, &u_ifreq64->ifr_ifru.ifru_data))
Arnd Bergmann7a229382009-11-06 23:00:29 -08003041 return -EFAULT;
3042
Arnd Bergmann6b960182009-11-06 23:10:54 -08003043 return dev_ioctl(net, cmd, u_ifreq64);
Arnd Bergmann7a229382009-11-06 23:00:29 -08003044}
3045
Arnd Bergmann6b960182009-11-06 23:10:54 -08003046static int dev_ifsioc(struct net *net, struct socket *sock,
3047 unsigned int cmd, struct compat_ifreq __user *uifr32)
Arnd Bergmann7a229382009-11-06 23:00:29 -08003048{
Arnd Bergmanna2116ed2009-11-11 03:39:40 +00003049 struct ifreq __user *uifr;
Arnd Bergmann7a229382009-11-06 23:00:29 -08003050 int err;
3051
Arnd Bergmanna2116ed2009-11-11 03:39:40 +00003052 uifr = compat_alloc_user_space(sizeof(*uifr));
3053 if (copy_in_user(uifr, uifr32, sizeof(*uifr32)))
3054 return -EFAULT;
3055
3056 err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr);
3057
Arnd Bergmann7a229382009-11-06 23:00:29 -08003058 if (!err) {
3059 switch (cmd) {
3060 case SIOCGIFFLAGS:
3061 case SIOCGIFMETRIC:
3062 case SIOCGIFMTU:
3063 case SIOCGIFMEM:
3064 case SIOCGIFHWADDR:
3065 case SIOCGIFINDEX:
3066 case SIOCGIFADDR:
3067 case SIOCGIFBRDADDR:
3068 case SIOCGIFDSTADDR:
3069 case SIOCGIFNETMASK:
Arnd Bergmannfab25322009-11-08 20:56:21 -08003070 case SIOCGIFPFLAGS:
Arnd Bergmann7a229382009-11-06 23:00:29 -08003071 case SIOCGIFTXQLEN:
Arnd Bergmannfab25322009-11-08 20:56:21 -08003072 case SIOCGMIIPHY:
3073 case SIOCGMIIREG:
Arnd Bergmanna2116ed2009-11-11 03:39:40 +00003074 if (copy_in_user(uifr32, uifr, sizeof(*uifr32)))
Arnd Bergmann7a229382009-11-06 23:00:29 -08003075 err = -EFAULT;
3076 break;
3077 }
3078 }
3079 return err;
3080}
3081
Arnd Bergmanna2116ed2009-11-11 03:39:40 +00003082static int compat_sioc_ifmap(struct net *net, unsigned int cmd,
3083 struct compat_ifreq __user *uifr32)
3084{
3085 struct ifreq ifr;
3086 struct compat_ifmap __user *uifmap32;
3087 mm_segment_t old_fs;
3088 int err;
3089
3090 uifmap32 = &uifr32->ifr_ifru.ifru_map;
3091 err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name));
Mathieu Desnoyers3ddc5b42013-09-11 14:23:18 -07003092 err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start);
3093 err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end);
3094 err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr);
3095 err |= get_user(ifr.ifr_map.irq, &uifmap32->irq);
3096 err |= get_user(ifr.ifr_map.dma, &uifmap32->dma);
3097 err |= get_user(ifr.ifr_map.port, &uifmap32->port);
Arnd Bergmanna2116ed2009-11-11 03:39:40 +00003098 if (err)
3099 return -EFAULT;
3100
3101 old_fs = get_fs();
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003102 set_fs(KERNEL_DS);
stephen hemmingerc3f52ae2011-02-23 09:06:48 +00003103 err = dev_ioctl(net, cmd, (void __user __force *)&ifr);
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003104 set_fs(old_fs);
Arnd Bergmanna2116ed2009-11-11 03:39:40 +00003105
3106 if (cmd == SIOCGIFMAP && !err) {
3107 err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name));
Mathieu Desnoyers3ddc5b42013-09-11 14:23:18 -07003108 err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start);
3109 err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end);
3110 err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr);
3111 err |= put_user(ifr.ifr_map.irq, &uifmap32->irq);
3112 err |= put_user(ifr.ifr_map.dma, &uifmap32->dma);
3113 err |= put_user(ifr.ifr_map.port, &uifmap32->port);
Arnd Bergmanna2116ed2009-11-11 03:39:40 +00003114 if (err)
3115 err = -EFAULT;
3116 }
3117 return err;
3118}
3119
Arnd Bergmann7a229382009-11-06 23:00:29 -08003120struct rtentry32 {
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003121 u32 rt_pad1;
Arnd Bergmann7a229382009-11-06 23:00:29 -08003122 struct sockaddr rt_dst; /* target address */
3123 struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */
3124 struct sockaddr rt_genmask; /* target network mask (IP) */
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003125 unsigned short rt_flags;
3126 short rt_pad2;
3127 u32 rt_pad3;
3128 unsigned char rt_tos;
3129 unsigned char rt_class;
3130 short rt_pad4;
3131 short rt_metric; /* +1 for binary compatibility! */
Arnd Bergmann7a229382009-11-06 23:00:29 -08003132 /* char * */ u32 rt_dev; /* forcing the device at add */
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003133 u32 rt_mtu; /* per route MTU/Window */
3134 u32 rt_window; /* Window clamping */
Arnd Bergmann7a229382009-11-06 23:00:29 -08003135 unsigned short rt_irtt; /* Initial RTT */
3136};
3137
3138struct in6_rtmsg32 {
3139 struct in6_addr rtmsg_dst;
3140 struct in6_addr rtmsg_src;
3141 struct in6_addr rtmsg_gateway;
3142 u32 rtmsg_type;
3143 u16 rtmsg_dst_len;
3144 u16 rtmsg_src_len;
3145 u32 rtmsg_metric;
3146 u32 rtmsg_info;
3147 u32 rtmsg_flags;
3148 s32 rtmsg_ifindex;
3149};
3150
Arnd Bergmann6b960182009-11-06 23:10:54 -08003151static int routing_ioctl(struct net *net, struct socket *sock,
3152 unsigned int cmd, void __user *argp)
Arnd Bergmann7a229382009-11-06 23:00:29 -08003153{
3154 int ret;
3155 void *r = NULL;
3156 struct in6_rtmsg r6;
3157 struct rtentry r4;
3158 char devname[16];
3159 u32 rtdev;
3160 mm_segment_t old_fs = get_fs();
3161
Arnd Bergmann6b960182009-11-06 23:10:54 -08003162 if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */
3163 struct in6_rtmsg32 __user *ur6 = argp;
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003164 ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst),
Arnd Bergmann7a229382009-11-06 23:00:29 -08003165 3 * sizeof(struct in6_addr));
Mathieu Desnoyers3ddc5b42013-09-11 14:23:18 -07003166 ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type));
3167 ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len));
3168 ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len));
3169 ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric));
3170 ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info));
3171 ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags));
3172 ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex));
Arnd Bergmann7a229382009-11-06 23:00:29 -08003173
3174 r = (void *) &r6;
3175 } else { /* ipv4 */
Arnd Bergmann6b960182009-11-06 23:10:54 -08003176 struct rtentry32 __user *ur4 = argp;
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003177 ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst),
Arnd Bergmann7a229382009-11-06 23:00:29 -08003178 3 * sizeof(struct sockaddr));
Mathieu Desnoyers3ddc5b42013-09-11 14:23:18 -07003179 ret |= get_user(r4.rt_flags, &(ur4->rt_flags));
3180 ret |= get_user(r4.rt_metric, &(ur4->rt_metric));
3181 ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu));
3182 ret |= get_user(r4.rt_window, &(ur4->rt_window));
3183 ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt));
3184 ret |= get_user(rtdev, &(ur4->rt_dev));
Arnd Bergmann7a229382009-11-06 23:00:29 -08003185 if (rtdev) {
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003186 ret |= copy_from_user(devname, compat_ptr(rtdev), 15);
stephen hemmingerc3f52ae2011-02-23 09:06:48 +00003187 r4.rt_dev = (char __user __force *)devname;
3188 devname[15] = 0;
Arnd Bergmann7a229382009-11-06 23:00:29 -08003189 } else
3190 r4.rt_dev = NULL;
3191
3192 r = (void *) &r4;
3193 }
3194
3195 if (ret) {
3196 ret = -EFAULT;
3197 goto out;
3198 }
3199
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003200 set_fs(KERNEL_DS);
Arnd Bergmann6b960182009-11-06 23:10:54 -08003201 ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r);
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003202 set_fs(old_fs);
Arnd Bergmann7a229382009-11-06 23:00:29 -08003203
3204out:
Arnd Bergmann7a229382009-11-06 23:00:29 -08003205 return ret;
3206}
3207
3208/* Since old style bridge ioctl's endup using SIOCDEVPRIVATE
3209 * for some operations; this forces use of the newer bridge-utils that
Lucas De Marchi25985ed2011-03-30 22:57:33 -03003210 * use compatible ioctls
Arnd Bergmann7a229382009-11-06 23:00:29 -08003211 */
Arnd Bergmann6b960182009-11-06 23:10:54 -08003212static int old_bridge_ioctl(compat_ulong_t __user *argp)
Arnd Bergmann7a229382009-11-06 23:00:29 -08003213{
Arnd Bergmann6b960182009-11-06 23:10:54 -08003214 compat_ulong_t tmp;
Arnd Bergmann7a229382009-11-06 23:00:29 -08003215
Arnd Bergmann6b960182009-11-06 23:10:54 -08003216 if (get_user(tmp, argp))
Arnd Bergmann7a229382009-11-06 23:00:29 -08003217 return -EFAULT;
3218 if (tmp == BRCTL_GET_VERSION)
3219 return BRCTL_VERSION + 1;
3220 return -EINVAL;
3221}
3222
Arnd Bergmann6b960182009-11-06 23:10:54 -08003223static int compat_sock_ioctl_trans(struct file *file, struct socket *sock,
3224 unsigned int cmd, unsigned long arg)
3225{
3226 void __user *argp = compat_ptr(arg);
3227 struct sock *sk = sock->sk;
3228 struct net *net = sock_net(sk);
Arnd Bergmann7a229382009-11-06 23:00:29 -08003229
Arnd Bergmann6b960182009-11-06 23:10:54 -08003230 if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))
Ben Hutchings590d4692013-11-18 17:04:13 +00003231 return compat_ifr_data_ioctl(net, cmd, argp);
Arnd Bergmann7a229382009-11-06 23:00:29 -08003232
Arnd Bergmann6b960182009-11-06 23:10:54 -08003233 switch (cmd) {
3234 case SIOCSIFBR:
3235 case SIOCGIFBR:
3236 return old_bridge_ioctl(argp);
3237 case SIOCGIFNAME:
3238 return dev_ifname32(net, argp);
3239 case SIOCGIFCONF:
3240 return dev_ifconf(net, argp);
3241 case SIOCETHTOOL:
3242 return ethtool_ioctl(net, argp);
Arnd Bergmann7a50a242009-11-08 20:57:03 -08003243 case SIOCWANDEV:
3244 return compat_siocwandev(net, argp);
Arnd Bergmanna2116ed2009-11-11 03:39:40 +00003245 case SIOCGIFMAP:
3246 case SIOCSIFMAP:
3247 return compat_sioc_ifmap(net, cmd, argp);
Arnd Bergmann6b960182009-11-06 23:10:54 -08003248 case SIOCBONDENSLAVE:
3249 case SIOCBONDRELEASE:
3250 case SIOCBONDSETHWADDR:
Arnd Bergmann6b960182009-11-06 23:10:54 -08003251 case SIOCBONDCHANGEACTIVE:
3252 return bond_ioctl(net, cmd, argp);
3253 case SIOCADDRT:
3254 case SIOCDELRT:
3255 return routing_ioctl(net, sock, cmd, argp);
3256 case SIOCGSTAMP:
3257 return do_siocgstamp(net, sock, cmd, argp);
3258 case SIOCGSTAMPNS:
3259 return do_siocgstampns(net, sock, cmd, argp);
Ben Hutchings590d4692013-11-18 17:04:13 +00003260 case SIOCBONDSLAVEINFOQUERY:
3261 case SIOCBONDINFOQUERY:
Arnd Bergmanna2116ed2009-11-11 03:39:40 +00003262 case SIOCSHWTSTAMP:
Ben Hutchingsfd468c72013-11-14 01:19:29 +00003263 case SIOCGHWTSTAMP:
Ben Hutchings590d4692013-11-18 17:04:13 +00003264 return compat_ifr_data_ioctl(net, cmd, argp);
Arnd Bergmann7a229382009-11-06 23:00:29 -08003265
Arnd Bergmann6b960182009-11-06 23:10:54 -08003266 case FIOSETOWN:
3267 case SIOCSPGRP:
3268 case FIOGETOWN:
3269 case SIOCGPGRP:
3270 case SIOCBRADDBR:
3271 case SIOCBRDELBR:
3272 case SIOCGIFVLAN:
3273 case SIOCSIFVLAN:
3274 case SIOCADDDLCI:
3275 case SIOCDELDLCI:
3276 return sock_ioctl(file, cmd, arg);
3277
3278 case SIOCGIFFLAGS:
3279 case SIOCSIFFLAGS:
3280 case SIOCGIFMETRIC:
3281 case SIOCSIFMETRIC:
3282 case SIOCGIFMTU:
3283 case SIOCSIFMTU:
3284 case SIOCGIFMEM:
3285 case SIOCSIFMEM:
3286 case SIOCGIFHWADDR:
3287 case SIOCSIFHWADDR:
3288 case SIOCADDMULTI:
3289 case SIOCDELMULTI:
3290 case SIOCGIFINDEX:
Arnd Bergmann6b960182009-11-06 23:10:54 -08003291 case SIOCGIFADDR:
3292 case SIOCSIFADDR:
3293 case SIOCSIFHWBROADCAST:
Arnd Bergmann6b960182009-11-06 23:10:54 -08003294 case SIOCDIFADDR:
Arnd Bergmann6b960182009-11-06 23:10:54 -08003295 case SIOCGIFBRDADDR:
3296 case SIOCSIFBRDADDR:
3297 case SIOCGIFDSTADDR:
3298 case SIOCSIFDSTADDR:
3299 case SIOCGIFNETMASK:
3300 case SIOCSIFNETMASK:
3301 case SIOCSIFPFLAGS:
3302 case SIOCGIFPFLAGS:
3303 case SIOCGIFTXQLEN:
3304 case SIOCSIFTXQLEN:
3305 case SIOCBRADDIF:
3306 case SIOCBRDELIF:
Arnd Bergmann9177efd2009-11-06 08:09:09 +00003307 case SIOCSIFNAME:
3308 case SIOCGMIIPHY:
3309 case SIOCGMIIREG:
3310 case SIOCSMIIREG:
Arnd Bergmann6b960182009-11-06 23:10:54 -08003311 return dev_ifsioc(net, sock, cmd, argp);
Arnd Bergmann9177efd2009-11-06 08:09:09 +00003312
Arnd Bergmann6b960182009-11-06 23:10:54 -08003313 case SIOCSARP:
3314 case SIOCGARP:
3315 case SIOCDARP:
Arnd Bergmann6b960182009-11-06 23:10:54 -08003316 case SIOCATMARK:
Arnd Bergmann9177efd2009-11-06 08:09:09 +00003317 return sock_do_ioctl(net, sock, cmd, arg);
3318 }
3319
Arnd Bergmann6b960182009-11-06 23:10:54 -08003320 return -ENOIOCTLCMD;
3321}
Arnd Bergmann7a229382009-11-06 23:00:29 -08003322
Eric Dumazet95c96172012-04-15 05:58:06 +00003323static long compat_sock_ioctl(struct file *file, unsigned int cmd,
Stephen Hemminger89bddce2006-09-01 00:19:31 -07003324 unsigned long arg)
Shaun Pereira89bbfc92006-03-21 23:58:08 -08003325{
3326 struct socket *sock = file->private_data;
3327 int ret = -ENOIOCTLCMD;
David S. Miller87de87d2008-06-03 09:14:03 -07003328 struct sock *sk;
3329 struct net *net;
3330
3331 sk = sock->sk;
3332 net = sock_net(sk);
Shaun Pereira89bbfc92006-03-21 23:58:08 -08003333
3334 if (sock->ops->compat_ioctl)
3335 ret = sock->ops->compat_ioctl(sock, cmd, arg);
3336
David S. Miller87de87d2008-06-03 09:14:03 -07003337 if (ret == -ENOIOCTLCMD &&
3338 (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST))
3339 ret = compat_wext_handle_ioctl(net, cmd, arg);
3340
Arnd Bergmann6b960182009-11-06 23:10:54 -08003341 if (ret == -ENOIOCTLCMD)
3342 ret = compat_sock_ioctl_trans(file, sock, cmd, arg);
3343
Shaun Pereira89bbfc92006-03-21 23:58:08 -08003344 return ret;
3345}
3346#endif
3347
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003348int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen)
3349{
3350 return sock->ops->bind(sock, addr, addrlen);
3351}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003352EXPORT_SYMBOL(kernel_bind);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003353
3354int kernel_listen(struct socket *sock, int backlog)
3355{
3356 return sock->ops->listen(sock, backlog);
3357}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003358EXPORT_SYMBOL(kernel_listen);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003359
3360int kernel_accept(struct socket *sock, struct socket **newsock, int flags)
3361{
3362 struct sock *sk = sock->sk;
3363 int err;
3364
3365 err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol,
3366 newsock);
3367 if (err < 0)
3368 goto done;
3369
3370 err = sock->ops->accept(sock, *newsock, flags);
3371 if (err < 0) {
3372 sock_release(*newsock);
Tony Battersbyfa8705b2007-10-10 21:09:04 -07003373 *newsock = NULL;
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003374 goto done;
3375 }
3376
3377 (*newsock)->ops = sock->ops;
Wei Yongjun1b085342008-12-18 19:35:10 -08003378 __module_get((*newsock)->ops->owner);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003379
3380done:
3381 return err;
3382}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003383EXPORT_SYMBOL(kernel_accept);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003384
3385int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen,
YOSHIFUJI Hideaki4768fbc2007-02-09 23:25:31 +09003386 int flags)
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003387{
3388 return sock->ops->connect(sock, addr, addrlen, flags);
3389}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003390EXPORT_SYMBOL(kernel_connect);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003391
3392int kernel_getsockname(struct socket *sock, struct sockaddr *addr,
3393 int *addrlen)
3394{
3395 return sock->ops->getname(sock, addr, addrlen, 0);
3396}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003397EXPORT_SYMBOL(kernel_getsockname);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003398
3399int kernel_getpeername(struct socket *sock, struct sockaddr *addr,
3400 int *addrlen)
3401{
3402 return sock->ops->getname(sock, addr, addrlen, 1);
3403}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003404EXPORT_SYMBOL(kernel_getpeername);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003405
3406int kernel_getsockopt(struct socket *sock, int level, int optname,
3407 char *optval, int *optlen)
3408{
3409 mm_segment_t oldfs = get_fs();
Namhyung Kimfb8621b2010-09-07 03:55:00 +00003410 char __user *uoptval;
3411 int __user *uoptlen;
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003412 int err;
3413
Namhyung Kimfb8621b2010-09-07 03:55:00 +00003414 uoptval = (char __user __force *) optval;
3415 uoptlen = (int __user __force *) optlen;
3416
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003417 set_fs(KERNEL_DS);
3418 if (level == SOL_SOCKET)
Namhyung Kimfb8621b2010-09-07 03:55:00 +00003419 err = sock_getsockopt(sock, level, optname, uoptval, uoptlen);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003420 else
Namhyung Kimfb8621b2010-09-07 03:55:00 +00003421 err = sock->ops->getsockopt(sock, level, optname, uoptval,
3422 uoptlen);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003423 set_fs(oldfs);
3424 return err;
3425}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003426EXPORT_SYMBOL(kernel_getsockopt);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003427
3428int kernel_setsockopt(struct socket *sock, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -07003429 char *optval, unsigned int optlen)
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003430{
3431 mm_segment_t oldfs = get_fs();
Namhyung Kimfb8621b2010-09-07 03:55:00 +00003432 char __user *uoptval;
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003433 int err;
3434
Namhyung Kimfb8621b2010-09-07 03:55:00 +00003435 uoptval = (char __user __force *) optval;
3436
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003437 set_fs(KERNEL_DS);
3438 if (level == SOL_SOCKET)
Namhyung Kimfb8621b2010-09-07 03:55:00 +00003439 err = sock_setsockopt(sock, level, optname, uoptval, optlen);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003440 else
Namhyung Kimfb8621b2010-09-07 03:55:00 +00003441 err = sock->ops->setsockopt(sock, level, optname, uoptval,
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003442 optlen);
3443 set_fs(oldfs);
3444 return err;
3445}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003446EXPORT_SYMBOL(kernel_setsockopt);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003447
3448int kernel_sendpage(struct socket *sock, struct page *page, int offset,
3449 size_t size, int flags)
3450{
3451 if (sock->ops->sendpage)
3452 return sock->ops->sendpage(sock, page, offset, size, flags);
3453
3454 return sock_no_sendpage(sock, page, offset, size, flags);
3455}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003456EXPORT_SYMBOL(kernel_sendpage);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003457
3458int kernel_sock_ioctl(struct socket *sock, int cmd, unsigned long arg)
3459{
3460 mm_segment_t oldfs = get_fs();
3461 int err;
3462
3463 set_fs(KERNEL_DS);
3464 err = sock->ops->ioctl(sock, cmd, arg);
3465 set_fs(oldfs);
3466
3467 return err;
3468}
Eric Dumazetc6d409c2010-06-03 20:03:40 -07003469EXPORT_SYMBOL(kernel_sock_ioctl);
Sridhar Samudralaac5a4882006-08-07 20:57:31 -07003470
Trond Myklebust91cf45f2007-11-12 18:10:39 -08003471int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how)
3472{
3473 return sock->ops->shutdown(sock, how);
3474}
Trond Myklebust91cf45f2007-11-12 18:10:39 -08003475EXPORT_SYMBOL(kernel_sock_shutdown);