Version:  2.0.40 2.2.26 2.4.37 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17 3.18

Linux/net/socket.c

  1 /*
  2  * NET          An implementation of the SOCKET network access protocol.
  3  *
  4  * Version:     @(#)socket.c    1.1.93  18/02/95
  5  *
  6  * Authors:     Orest Zborowski, <obz@Kodak.COM>
  7  *              Ross Biro
  8  *              Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
  9  *
 10  * Fixes:
 11  *              Anonymous       :       NOTSOCK/BADF cleanup. Error fix in
 12  *                                      shutdown()
 13  *              Alan Cox        :       verify_area() fixes
 14  *              Alan Cox        :       Removed DDI
 15  *              Jonathan Kamens :       SOCK_DGRAM reconnect bug
 16  *              Alan Cox        :       Moved a load of checks to the very
 17  *                                      top level.
 18  *              Alan Cox        :       Move address structures to/from user
 19  *                                      mode above the protocol layers.
 20  *              Rob Janssen     :       Allow 0 length sends.
 21  *              Alan Cox        :       Asynchronous I/O support (cribbed from the
 22  *                                      tty drivers).
 23  *              Niibe Yutaka    :       Asynchronous I/O for writes (4.4BSD style)
 24  *              Jeff Uphoff     :       Made max number of sockets command-line
 25  *                                      configurable.
 26  *              Matti Aarnio    :       Made the number of sockets dynamic,
 27  *                                      to be allocated when needed, and mr.
 28  *                                      Uphoff's max is used as max to be
 29  *                                      allowed to allocate.
 30  *              Linus           :       Argh. removed all the socket allocation
 31  *                                      altogether: it's in the inode now.
 32  *              Alan Cox        :       Made sock_alloc()/sock_release() public
 33  *                                      for NetROM and future kernel nfsd type
 34  *                                      stuff.
 35  *              Alan Cox        :       sendmsg/recvmsg basics.
 36  *              Tom Dyas        :       Export net symbols.
 37  *              Marcin Dalecki  :       Fixed problems with CONFIG_NET="n".
 38  *              Alan Cox        :       Added thread locking to sys_* calls
 39  *                                      for sockets. May have errors at the
 40  *                                      moment.
 41  *              Kevin Buhr      :       Fixed the dumb errors in the above.
 42  *              Andi Kleen      :       Some small cleanups, optimizations,
 43  *                                      and fixed a copy_from_user() bug.
 44  *              Tigran Aivazian :       sys_send(args) calls sys_sendto(args, NULL, 0)
 45  *              Tigran Aivazian :       Made listen(2) backlog sanity checks
 46  *                                      protocol-independent
 47  *
 48  *
 49  *              This program is free software; you can redistribute it and/or
 50  *              modify it under the terms of the GNU General Public License
 51  *              as published by the Free Software Foundation; either version
 52  *              2 of the License, or (at your option) any later version.
 53  *
 54  *
 55  *      This module is effectively the top level interface to the BSD socket
 56  *      paradigm.
 57  *
 58  *      Based upon Swansea University Computer Society NET3.039
 59  */
 60 
 61 #include <linux/mm.h>
 62 #include <linux/socket.h>
 63 #include <linux/file.h>
 64 #include <linux/net.h>
 65 #include <linux/interrupt.h>
 66 #include <linux/thread_info.h>
 67 #include <linux/rcupdate.h>
 68 #include <linux/netdevice.h>
 69 #include <linux/proc_fs.h>
 70 #include <linux/seq_file.h>
 71 #include <linux/mutex.h>
 72 #include <linux/if_bridge.h>
 73 #include <linux/if_frad.h>
 74 #include <linux/if_vlan.h>
 75 #include <linux/ptp_classify.h>
 76 #include <linux/init.h>
 77 #include <linux/poll.h>
 78 #include <linux/cache.h>
 79 #include <linux/module.h>
 80 #include <linux/highmem.h>
 81 #include <linux/mount.h>
 82 #include <linux/security.h>
 83 #include <linux/syscalls.h>
 84 #include <linux/compat.h>
 85 #include <linux/kmod.h>
 86 #include <linux/audit.h>
 87 #include <linux/wireless.h>
 88 #include <linux/nsproxy.h>
 89 #include <linux/magic.h>
 90 #include <linux/slab.h>
 91 #include <linux/xattr.h>
 92 
 93 #include <asm/uaccess.h>
 94 #include <asm/unistd.h>
 95 
 96 #include <net/compat.h>
 97 #include <net/wext.h>
 98 #include <net/cls_cgroup.h>
 99 
100 #include <net/sock.h>
101 #include <linux/netfilter.h>
102 
103 #include <linux/if_tun.h>
104 #include <linux/ipv6_route.h>
105 #include <linux/route.h>
106 #include <linux/sockios.h>
107 #include <linux/atalk.h>
108 #include <net/busy_poll.h>
109 #include <linux/errqueue.h>
110 
111 #ifdef CONFIG_NET_RX_BUSY_POLL
112 unsigned int sysctl_net_busy_read __read_mostly;
113 unsigned int sysctl_net_busy_poll __read_mostly;
114 #endif
115 
116 static int sock_no_open(struct inode *irrelevant, struct file *dontcare);
117 static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov,
118                          unsigned long nr_segs, loff_t pos);
119 static ssize_t sock_aio_write(struct kiocb *iocb, const struct iovec *iov,
120                           unsigned long nr_segs, loff_t pos);
121 static int sock_mmap(struct file *file, struct vm_area_struct *vma);
122 
123 static int sock_close(struct inode *inode, struct file *file);
124 static unsigned int sock_poll(struct file *file,
125                               struct poll_table_struct *wait);
126 static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
127 #ifdef CONFIG_COMPAT
128 static long compat_sock_ioctl(struct file *file,
129                               unsigned int cmd, unsigned long arg);
130 #endif
131 static int sock_fasync(int fd, struct file *filp, int on);
132 static ssize_t sock_sendpage(struct file *file, struct page *page,
133                              int offset, size_t size, loff_t *ppos, int more);
134 static ssize_t sock_splice_read(struct file *file, loff_t *ppos,
135                                 struct pipe_inode_info *pipe, size_t len,
136                                 unsigned int flags);
137 
138 /*
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 
143 static const struct file_operations socket_file_ops = {
144         .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,
150 #ifdef CONFIG_COMPAT
151         .compat_ioctl = compat_sock_ioctl,
152 #endif
153         .mmap =         sock_mmap,
154         .open =         sock_no_open,   /* special open code to disallow open via /proc */
155         .release =      sock_close,
156         .fasync =       sock_fasync,
157         .sendpage =     sock_sendpage,
158         .splice_write = generic_splice_sendpage,
159         .splice_read =  sock_splice_read,
160 };
161 
162 /*
163  *      The protocol list. Each protocol is registered in here.
164  */
165 
166 static DEFINE_SPINLOCK(net_family_lock);
167 static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly;
168 
169 /*
170  *      Statistics counters of the socket lists
171  */
172 
173 static DEFINE_PER_CPU(int, sockets_in_use);
174 
175 /*
176  * Support routines.
177  * Move socket addresses back and forth across the kernel/user
178  * divide and look after the messy bits.
179  */
180 
181 /**
182  *      move_addr_to_kernel     -       copy a socket address into kernel space
183  *      @uaddr: Address in user space
184  *      @kaddr: Address in kernel space
185  *      @ulen: Length in user space
186  *
187  *      The address is copied into kernel space. If the provided address is
188  *      too long an error code of -EINVAL is returned. If the copy gives
189  *      invalid addresses -EFAULT is returned. On a success 0 is returned.
190  */
191 
192 int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr)
193 {
194         if (ulen < 0 || ulen > sizeof(struct sockaddr_storage))
195                 return -EINVAL;
196         if (ulen == 0)
197                 return 0;
198         if (copy_from_user(kaddr, uaddr, ulen))
199                 return -EFAULT;
200         return audit_sockaddr(ulen, kaddr);
201 }
202 
203 /**
204  *      move_addr_to_user       -       copy an address to user space
205  *      @kaddr: kernel space address
206  *      @klen: length of address in kernel
207  *      @uaddr: user space address
208  *      @ulen: pointer to user length field
209  *
210  *      The value pointed to by ulen on entry is the buffer length available.
211  *      This is overwritten with the buffer space used. -EINVAL is returned
212  *      if an overlong buffer is specified or a negative buffer size. -EFAULT
213  *      is returned if either the buffer or the length field are not
214  *      accessible.
215  *      After copying the data up to the limit the user specifies, the true
216  *      length of the data is written over the length limit the user
217  *      specified. Zero is returned for a success.
218  */
219 
220 static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen,
221                              void __user *uaddr, int __user *ulen)
222 {
223         int err;
224         int len;
225 
226         BUG_ON(klen > sizeof(struct sockaddr_storage));
227         err = get_user(len, ulen);
228         if (err)
229                 return err;
230         if (len > klen)
231                 len = klen;
232         if (len < 0)
233                 return -EINVAL;
234         if (len) {
235                 if (audit_sockaddr(klen, kaddr))
236                         return -ENOMEM;
237                 if (copy_to_user(uaddr, kaddr, len))
238                         return -EFAULT;
239         }
240         /*
241          *      "fromlen shall refer to the value before truncation.."
242          *                      1003.1g
243          */
244         return __put_user(klen, ulen);
245 }
246 
247 static struct kmem_cache *sock_inode_cachep __read_mostly;
248 
249 static struct inode *sock_alloc_inode(struct super_block *sb)
250 {
251         struct socket_alloc *ei;
252         struct socket_wq *wq;
253 
254         ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL);
255         if (!ei)
256                 return NULL;
257         wq = kmalloc(sizeof(*wq), GFP_KERNEL);
258         if (!wq) {
259                 kmem_cache_free(sock_inode_cachep, ei);
260                 return NULL;
261         }
262         init_waitqueue_head(&wq->wait);
263         wq->fasync_list = NULL;
264         RCU_INIT_POINTER(ei->socket.wq, wq);
265 
266         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;
271 
272         return &ei->vfs_inode;
273 }
274 
275 static void sock_destroy_inode(struct inode *inode)
276 {
277         struct socket_alloc *ei;
278         struct socket_wq *wq;
279 
280         ei = container_of(inode, struct socket_alloc, vfs_inode);
281         wq = rcu_dereference_protected(ei->socket.wq, 1);
282         kfree_rcu(wq, rcu);
283         kmem_cache_free(sock_inode_cachep, ei);
284 }
285 
286 static void init_once(void *foo)
287 {
288         struct socket_alloc *ei = (struct socket_alloc *)foo;
289 
290         inode_init_once(&ei->vfs_inode);
291 }
292 
293 static int init_inodecache(void)
294 {
295         sock_inode_cachep = kmem_cache_create("sock_inode_cache",
296                                               sizeof(struct socket_alloc),
297                                               0,
298                                               (SLAB_HWCACHE_ALIGN |
299                                                SLAB_RECLAIM_ACCOUNT |
300                                                SLAB_MEM_SPREAD),
301                                               init_once);
302         if (sock_inode_cachep == NULL)
303                 return -ENOMEM;
304         return 0;
305 }
306 
307 static const struct super_operations sockfs_ops = {
308         .alloc_inode    = sock_alloc_inode,
309         .destroy_inode  = sock_destroy_inode,
310         .statfs         = simple_statfs,
311 };
312 
313 /*
314  * sockfs_dname() is called from d_path().
315  */
316 static 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 
322 static const struct dentry_operations sockfs_dentry_operations = {
323         .d_dname  = sockfs_dname,
324 };
325 
326 static 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 
333 static struct vfsmount *sock_mnt __read_mostly;
334 
335 static struct file_system_type sock_fs_type = {
336         .name =         "sockfs",
337         .mount =        sockfs_mount,
338         .kill_sb =      kill_anon_super,
339 };
340 
341 /*
342  *      Obtains the first available file descriptor and sets it up for use.
343  *
344  *      These functions create file structures and maps them to fd space
345  *      of the current process. On success it returns file descriptor
346  *      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 
358 struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname)
359 {
360         struct qstr name = { .name = "" };
361         struct path path;
362         struct file *file;
363 
364         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         }
371         path.dentry = d_alloc_pseudo(sock_mnt->mnt_sb, &name);
372         if (unlikely(!path.dentry))
373                 return ERR_PTR(-ENOMEM);
374         path.mnt = mntget(sock_mnt);
375 
376         d_instantiate(path.dentry, SOCK_INODE(sock));
377         SOCK_INODE(sock)->i_fop = &socket_file_ops;
378 
379         file = alloc_file(&path, FMODE_READ | FMODE_WRITE,
380                   &socket_file_ops);
381         if (unlikely(IS_ERR(file))) {
382                 /* drop dentry, keep inode */
383                 ihold(path.dentry->d_inode);
384                 path_put(&path);
385                 return file;
386         }
387 
388         sock->file = file;
389         file->f_flags = O_RDWR | (flags & O_NONBLOCK);
390         file->private_data = sock;
391         return file;
392 }
393 EXPORT_SYMBOL(sock_alloc_file);
394 
395 static int sock_map_fd(struct socket *sock, int flags)
396 {
397         struct file *newfile;
398         int fd = get_unused_fd_flags(flags);
399         if (unlikely(fd < 0))
400                 return fd;
401 
402         newfile = sock_alloc_file(sock, flags, NULL);
403         if (likely(!IS_ERR(newfile))) {
404                 fd_install(fd, newfile);
405                 return fd;
406         }
407 
408         put_unused_fd(fd);
409         return PTR_ERR(newfile);
410 }
411 
412 struct socket *sock_from_file(struct file *file, int *err)
413 {
414         if (file->f_op == &socket_file_ops)
415                 return file->private_data;      /* set in sock_map_fd */
416 
417         *err = -ENOTSOCK;
418         return NULL;
419 }
420 EXPORT_SYMBOL(sock_from_file);
421 
422 /**
423  *      sockfd_lookup - Go from a file number to its socket slot
424  *      @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 
435 struct socket *sockfd_lookup(int fd, int *err)
436 {
437         struct file *file;
438         struct socket *sock;
439 
440         file = fget(fd);
441         if (!file) {
442                 *err = -EBADF;
443                 return NULL;
444         }
445 
446         sock = sock_from_file(file, err);
447         if (!sock)
448                 fput(file);
449         return sock;
450 }
451 EXPORT_SYMBOL(sockfd_lookup);
452 
453 static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed)
454 {
455         struct fd f = fdget(fd);
456         struct socket *sock;
457 
458         *err = -EBADF;
459         if (f.file) {
460                 sock = sock_from_file(f.file, err);
461                 if (likely(sock)) {
462                         *fput_needed = f.flags;
463                         return sock;
464                 }
465                 fdput(f);
466         }
467         return NULL;
468 }
469 
470 #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)
473 static 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 
495 out:
496         return error;
497 }
498 
499 static 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 
527 static const struct inode_operations sockfs_inode_ops = {
528         .getxattr = sockfs_getxattr,
529         .listxattr = sockfs_listxattr,
530 };
531 
532 /**
533  *      sock_alloc      -       allocate a socket
534  *
535  *      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 
540 static struct socket *sock_alloc(void)
541 {
542         struct inode *inode;
543         struct socket *sock;
544 
545         inode = new_inode_pseudo(sock_mnt->mnt_sb);
546         if (!inode)
547                 return NULL;
548 
549         sock = SOCKET_I(inode);
550 
551         kmemcheck_annotate_bitfield(sock, type);
552         inode->i_ino = get_next_ino();
553         inode->i_mode = S_IFSOCK | S_IRWXUGO;
554         inode->i_uid = current_fsuid();
555         inode->i_gid = current_fsgid();
556         inode->i_op = &sockfs_inode_ops;
557 
558         this_cpu_add(sockets_in_use, 1);
559         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  */
567 
568 static int sock_no_open(struct inode *irrelevant, struct file *dontcare)
569 {
570         return -ENXIO;
571 }
572 
573 const struct file_operations bad_sock_fops = {
574         .owner = THIS_MODULE,
575         .open = sock_no_open,
576         .llseek = noop_llseek,
577 };
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
585  *      an inode not a file.
586  */
587 
588 void 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 
598         if (rcu_dereference_protected(sock->wq, 1)->fasync_list)
599                 pr_err("%s: fasync list not empty!\n", __func__);
600 
601         if (test_bit(SOCK_EXTERNALLY_ALLOCATED, &sock->flags))
602                 return;
603 
604         this_cpu_sub(sockets_in_use, 1);
605         if (!sock->file) {
606                 iput(SOCK_INODE(sock));
607                 return;
608         }
609         sock->file = NULL;
610 }
611 EXPORT_SYMBOL(sock_release);
612 
613 void __sock_tx_timestamp(const struct sock *sk, __u8 *tx_flags)
614 {
615         u8 flags = *tx_flags;
616 
617         if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_HARDWARE)
618                 flags |= SKBTX_HW_TSTAMP;
619 
620         if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_SOFTWARE)
621                 flags |= SKBTX_SW_TSTAMP;
622 
623         if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_SCHED)
624                 flags |= SKBTX_SCHED_TSTAMP;
625 
626         if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_ACK)
627                 flags |= SKBTX_ACK_TSTAMP;
628 
629         *tx_flags = flags;
630 }
631 EXPORT_SYMBOL(__sock_tx_timestamp);
632 
633 static inline int __sock_sendmsg_nosec(struct kiocb *iocb, struct socket *sock,
634                                        struct msghdr *msg, size_t size)
635 {
636         struct sock_iocb *si = kiocb_to_siocb(iocb);
637 
638         si->sock = sock;
639         si->scm = NULL;
640         si->msg = msg;
641         si->size = size;
642 
643         return sock->ops->sendmsg(iocb, sock, msg, size);
644 }
645 
646 static 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 
654 int 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 }
667 EXPORT_SYMBOL(sock_sendmsg);
668 
669 static int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg, size_t size)
670 {
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 
683 int 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          */
694         msg->msg_iov = (struct iovec *)vec;
695         msg->msg_iovlen = num;
696         result = sock_sendmsg(sock, msg, size);
697         set_fs(oldfs);
698         return result;
699 }
700 EXPORT_SYMBOL(kernel_sendmsg);
701 
702 /*
703  * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP)
704  */
705 void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk,
706         struct sk_buff *skb)
707 {
708         int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP);
709         struct scm_timestamping tss;
710         int empty = 1;
711         struct skb_shared_hwtstamps *shhwtstamps =
712                 skb_hwtstamps(skb);
713 
714         /* 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 {
726                         struct timespec ts;
727                         skb_get_timestampns(skb, &ts);
728                         put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPNS,
729                                  sizeof(ts), &ts);
730                 }
731         }
732 
733         memset(&tss, 0, sizeof(tss));
734         if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) &&
735             ktime_to_timespec_cond(skb->tstamp, tss.ts + 0))
736                 empty = 0;
737         if (shhwtstamps &&
738             (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) &&
739             ktime_to_timespec_cond(shhwtstamps->hwtstamp, tss.ts + 2))
740                 empty = 0;
741         if (!empty)
742                 put_cmsg(msg, SOL_SOCKET,
743                          SCM_TIMESTAMPING, sizeof(tss), &tss);
744 }
745 EXPORT_SYMBOL_GPL(__sock_recv_timestamp);
746 
747 void __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 }
761 EXPORT_SYMBOL_GPL(__sock_recv_wifi_status);
762 
763 static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk,
764                                    struct sk_buff *skb)
765 {
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 
771 void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk,
772         struct sk_buff *skb)
773 {
774         sock_recv_timestamp(msg, sk, skb);
775         sock_recv_drops(msg, sk, skb);
776 }
777 EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops);
778 
779 static inline int __sock_recvmsg_nosec(struct kiocb *iocb, struct socket *sock,
780                                        struct msghdr *msg, size_t size, int flags)
781 {
782         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 
790         return sock->ops->recvmsg(iocb, sock, msg, size, flags);
791 }
792 
793 static 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 
801 int sock_recvmsg(struct socket *sock, struct msghdr *msg,
802                  size_t size, int flags)
803 {
804         struct kiocb iocb;
805         struct sock_iocb siocb;
806         int ret;
807 
808         init_sync_kiocb(&iocb, NULL);
809         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 }
815 EXPORT_SYMBOL(sock_recvmsg);
816 
817 static 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 
832 /**
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  */
847 int kernel_recvmsg(struct socket *sock, struct msghdr *msg,
848                    struct kvec *vec, size_t num, size_t size, int flags)
849 {
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          */
858         msg->msg_iov = (struct iovec *)vec, msg->msg_iovlen = num;
859         result = sock_recvmsg(sock, msg, size, flags);
860         set_fs(oldfs);
861         return result;
862 }
863 EXPORT_SYMBOL(kernel_recvmsg);
864 
865 static ssize_t sock_sendpage(struct file *file, struct page *page,
866                              int offset, size_t size, loff_t *ppos, int more)
867 {
868         struct socket *sock;
869         int flags;
870 
871         sock = file->private_data;
872 
873         flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0;
874         /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */
875         flags |= more;
876 
877         return kernel_sendpage(sock, page, offset, size, flags);
878 }
879 
880 static ssize_t sock_splice_read(struct file *file, loff_t *ppos,
881                                 struct pipe_inode_info *pipe, size_t len,
882                                 unsigned int flags)
883 {
884         struct socket *sock = file->private_data;
885 
886         if (unlikely(!sock->ops->splice_read))
887                 return -EINVAL;
888 
889         return sock->ops->splice_read(sock, ppos, pipe, len, flags);
890 }
891 
892 static struct sock_iocb *alloc_sock_iocb(struct kiocb *iocb,
893                                          struct sock_iocb *siocb)
894 {
895         if (!is_sync_kiocb(iocb))
896                 BUG();
897 
898         siocb->kiocb = iocb;
899         iocb->private = siocb;
900         return siocb;
901 }
902 
903 static ssize_t do_sock_read(struct msghdr *msg, struct kiocb *iocb,
904                 struct file *file, const struct iovec *iov,
905                 unsigned long nr_segs)
906 {
907         struct socket *sock = file->private_data;
908         size_t size = 0;
909         int i;
910 
911         for (i = 0; i < nr_segs; i++)
912                 size += iov[i].iov_len;
913 
914         msg->msg_name = NULL;
915         msg->msg_namelen = 0;
916         msg->msg_control = NULL;
917         msg->msg_controllen = 0;
918         msg->msg_iov = (struct iovec *)iov;
919         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 
925 static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov,
926                                 unsigned long nr_segs, loff_t pos)
927 {
928         struct sock_iocb siocb, *x;
929 
930         if (pos != 0)
931                 return -ESPIPE;
932 
933         if (iocb->ki_nbytes == 0)       /* Match SYS5 behaviour */
934                 return 0;
935 
936 
937         x = alloc_sock_iocb(iocb, &siocb);
938         if (!x)
939                 return -ENOMEM;
940         return do_sock_read(&x->async_msg, iocb, iocb->ki_filp, iov, nr_segs);
941 }
942 
943 static ssize_t do_sock_write(struct msghdr *msg, struct kiocb *iocb,
944                         struct file *file, const struct iovec *iov,
945                         unsigned long nr_segs)
946 {
947         struct socket *sock = file->private_data;
948         size_t size = 0;
949         int i;
950 
951         for (i = 0; i < nr_segs; i++)
952                 size += iov[i].iov_len;
953 
954         msg->msg_name = NULL;
955         msg->msg_namelen = 0;
956         msg->msg_control = NULL;
957         msg->msg_controllen = 0;
958         msg->msg_iov = (struct iovec *)iov;
959         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 
967 static ssize_t sock_aio_write(struct kiocb *iocb, const struct iovec *iov,
968                           unsigned long nr_segs, loff_t pos)
969 {
970         struct sock_iocb siocb, *x;
971 
972         if (pos != 0)
973                 return -ESPIPE;
974 
975         x = alloc_sock_iocb(iocb, &siocb);
976         if (!x)
977                 return -ENOMEM;
978 
979         return do_sock_write(&x->async_msg, iocb, iocb->ki_filp, iov, nr_segs);
980 }
981 
982 /*
983  * Atomic setting of ioctl hooks to avoid race
984  * with module unload.
985  */
986 
987 static DEFINE_MUTEX(br_ioctl_mutex);
988 static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg);
989 
990 void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *))
991 {
992         mutex_lock(&br_ioctl_mutex);
993         br_ioctl_hook = hook;
994         mutex_unlock(&br_ioctl_mutex);
995 }
996 EXPORT_SYMBOL(brioctl_set);
997 
998 static DEFINE_MUTEX(vlan_ioctl_mutex);
999 static int (*vlan_ioctl_hook) (struct net *, void __user *arg);
1000 
1001 void vlan_ioctl_set(int (*hook) (struct net *, void __user *))
1002 {
1003         mutex_lock(&vlan_ioctl_mutex);
1004         vlan_ioctl_hook = hook;
1005         mutex_unlock(&vlan_ioctl_mutex);
1006 }
1007 EXPORT_SYMBOL(vlan_ioctl_set);
1008 
1009 static DEFINE_MUTEX(dlci_ioctl_mutex);
1010 static int (*dlci_ioctl_hook) (unsigned int, void __user *);
1011 
1012 void dlci_ioctl_set(int (*hook) (unsigned int, void __user *))
1013 {
1014         mutex_lock(&dlci_ioctl_mutex);
1015         dlci_ioctl_hook = hook;
1016         mutex_unlock(&dlci_ioctl_mutex);
1017 }
1018 EXPORT_SYMBOL(dlci_ioctl_set);
1019 
1020 static 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 
1038 /*
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 
1043 static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg)
1044 {
1045         struct socket *sock;
1046         struct sock *sk;
1047         void __user *argp = (void __user *)arg;
1048         int pid, err;
1049         struct net *net;
1050 
1051         sock = file->private_data;
1052         sk = sock->sk;
1053         net = sock_net(sk);
1054         if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) {
1055                 err = dev_ioctl(net, cmd, argp);
1056         } else
1057 #ifdef CONFIG_WEXT_CORE
1058         if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) {
1059                 err = dev_ioctl(net, cmd, argp);
1060         } else
1061 #endif
1062                 switch (cmd) {
1063                 case FIOSETOWN:
1064                 case SIOCSPGRP:
1065                         err = -EFAULT;
1066                         if (get_user(pid, (int __user *)argp))
1067                                 break;
1068                         f_setown(sock->file, pid, 1);
1069                         err = 0;
1070                         break;
1071                 case FIOGETOWN:
1072                 case SIOCGPGRP:
1073                         err = put_user(f_getown(sock->file),
1074                                        (int __user *)argp);
1075                         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 
1084                         mutex_lock(&br_ioctl_mutex);
1085                         if (br_ioctl_hook)
1086                                 err = br_ioctl_hook(net, cmd, argp);
1087                         mutex_unlock(&br_ioctl_mutex);
1088                         break;
1089                 case SIOCGIFVLAN:
1090                 case SIOCSIFVLAN:
1091                         err = -ENOPKG;
1092                         if (!vlan_ioctl_hook)
1093                                 request_module("8021q");
1094 
1095                         mutex_lock(&vlan_ioctl_mutex);
1096                         if (vlan_ioctl_hook)
1097                                 err = vlan_ioctl_hook(net, argp);
1098                         mutex_unlock(&vlan_ioctl_mutex);
1099                         break;
1100                 case SIOCADDDLCI:
1101                 case SIOCDELDLCI:
1102                         err = -ENOPKG;
1103                         if (!dlci_ioctl_hook)
1104                                 request_module("dlci");
1105 
1106                         mutex_lock(&dlci_ioctl_mutex);
1107                         if (dlci_ioctl_hook)
1108                                 err = dlci_ioctl_hook(cmd, argp);
1109                         mutex_unlock(&dlci_ioctl_mutex);
1110                         break;
1111                 default:
1112                         err = sock_do_ioctl(net, sock, cmd, arg);
1113                         break;
1114                 }
1115         return err;
1116 }
1117 
1118 int sock_create_lite(int family, int type, int protocol, struct socket **res)
1119 {
1120         int err;
1121         struct socket *sock = NULL;
1122 
1123         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 
1133         sock->type = type;
1134         err = security_socket_post_create(sock, family, type, protocol, 1);
1135         if (err)
1136                 goto out_release;
1137 
1138 out:
1139         *res = sock;
1140         return err;
1141 out_release:
1142         sock_release(sock);
1143         sock = NULL;
1144         goto out;
1145 }
1146 EXPORT_SYMBOL(sock_create_lite);
1147 
1148 /* No kernel lock held - perfect */
1149 static unsigned int sock_poll(struct file *file, poll_table *wait)
1150 {
1151         unsigned int busy_flag = 0;
1152         struct socket *sock;
1153 
1154         /*
1155          *      We can't return errors to poll, so it's either yes or no.
1156          */
1157         sock = file->private_data;
1158 
1159         if (sk_can_busy_loop(sock->sk)) {
1160                 /* this socket can poll_ll so tell the system call */
1161                 busy_flag = POLL_BUSY_LOOP;
1162 
1163                 /* once, only if requested by syscall */
1164                 if (wait && (wait->_key & POLL_BUSY_LOOP))
1165                         sk_busy_loop(sock->sk, 1);
1166         }
1167 
1168         return busy_flag | sock->ops->poll(file, sock, wait);
1169 }
1170 
1171 static int sock_mmap(struct file *file, struct vm_area_struct *vma)
1172 {
1173         struct socket *sock = file->private_data;
1174 
1175         return sock->ops->mmap(file, sock, vma);
1176 }
1177 
1178 static int sock_close(struct inode *inode, struct file *filp)
1179 {
1180         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)
1192  *         or under socket lock
1193  */
1194 
1195 static int sock_fasync(int fd, struct file *filp, int on)
1196 {
1197         struct socket *sock = filp->private_data;
1198         struct sock *sk = sock->sk;
1199         struct socket_wq *wq;
1200 
1201         if (sk == NULL)
1202                 return -EINVAL;
1203 
1204         lock_sock(sk);
1205         wq = rcu_dereference_protected(sock->wq, sock_owned_by_user(sk));
1206         fasync_helper(fd, filp, on, &wq->fasync_list);
1207 
1208         if (!wq->fasync_list)
1209                 sock_reset_flag(sk, SOCK_FASYNC);
1210         else
1211                 sock_set_flag(sk, SOCK_FASYNC);
1212 
1213         release_sock(sk);
1214         return 0;
1215 }
1216 
1217 /* This function may be called only under socket lock or callback_lock or rcu_lock */
1218 
1219 int sock_wake_async(struct socket *sock, int how, int band)
1220 {
1221         struct socket_wq *wq;
1222 
1223         if (!sock)
1224                 return -1;
1225         rcu_read_lock();
1226         wq = rcu_dereference(sock->wq);
1227         if (!wq || !wq->fasync_list) {
1228                 rcu_read_unlock();
1229                 return -1;
1230         }
1231         switch (how) {
1232         case SOCK_WAKE_WAITD:
1233                 if (test_bit(SOCK_ASYNC_WAITDATA, &sock->flags))
1234                         break;
1235                 goto call_kill;
1236         case SOCK_WAKE_SPACE:
1237                 if (!test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags))
1238                         break;
1239                 /* fall through */
1240         case SOCK_WAKE_IO:
1241 call_kill:
1242                 kill_fasync(&wq->fasync_list, SIGIO, band);
1243                 break;
1244         case SOCK_WAKE_URG:
1245                 kill_fasync(&wq->fasync_list, SIGURG, band);
1246         }
1247         rcu_read_unlock();
1248         return 0;
1249 }
1250 EXPORT_SYMBOL(sock_wake_async);
1251 
1252 int __sock_create(struct net *net, int family, int type, int protocol,
1253                          struct socket **res, int kern)
1254 {
1255         int err;
1256         struct socket *sock;
1257         const struct net_proto_family *pf;
1258 
1259         /*
1260          *      Check protocol is in range
1261          */
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) {
1273                 static int warned;
1274                 if (!warned) {
1275                         warned = 1;
1276                         pr_info("%s uses obsolete (PF_INET,SOCK_PACKET)\n",
1277                                 current->comm);
1278                 }
1279                 family = PF_PACKET;
1280         }
1281 
1282         err = security_socket_create(family, type, protocol, kern);
1283         if (err)
1284                 return err;
1285 
1286         /*
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) {
1293                 net_warn_ratelimited("socket: no more sockets\n");
1294                 return -ENFILE; /* Not exactly a match, but its the
1295                                    closest posix thing */
1296         }
1297 
1298         sock->type = type;
1299 
1300 #ifdef CONFIG_MODULES
1301         /* 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
1304          * requested real, full-featured networking support upon configuration.
1305          * Otherwise module support will break!
1306          */
1307         if (rcu_access_pointer(net_families[family]) == NULL)
1308                 request_module("net-pf-%d", family);
1309 #endif
1310 
1311         rcu_read_lock();
1312         pf = rcu_dereference(net_families[family]);
1313         err = -EAFNOSUPPORT;
1314         if (!pf)
1315                 goto out_release;
1316 
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          */
1321         if (!try_module_get(pf->owner))
1322                 goto out_release;
1323 
1324         /* Now protected by module ref count */
1325         rcu_read_unlock();
1326 
1327         err = pf->create(net, sock, protocol, kern);
1328         if (err < 0)
1329                 goto out_module_put;
1330 
1331         /*
1332          * Now to bump the refcnt of the [loadable] module that owns this
1333          * socket at sock_release time we decrement its refcnt.
1334          */
1335         if (!try_module_get(sock->ops->owner))
1336                 goto out_module_busy;
1337 
1338         /*
1339          * Now that we're done with the ->create function, the [loadable]
1340          * module can have its refcnt decremented
1341          */
1342         module_put(pf->owner);
1343         err = security_socket_post_create(sock, family, type, protocol, kern);
1344         if (err)
1345                 goto out_sock_release;
1346         *res = sock;
1347 
1348         return 0;
1349 
1350 out_module_busy:
1351         err = -EAFNOSUPPORT;
1352 out_module_put:
1353         sock->ops = NULL;
1354         module_put(pf->owner);
1355 out_sock_release:
1356         sock_release(sock);
1357         return err;
1358 
1359 out_release:
1360         rcu_read_unlock();
1361         goto out_sock_release;
1362 }
1363 EXPORT_SYMBOL(__sock_create);
1364 
1365 int sock_create(int family, int type, int protocol, struct socket **res)
1366 {
1367         return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0);
1368 }
1369 EXPORT_SYMBOL(sock_create);
1370 
1371 int sock_create_kern(int family, int type, int protocol, struct socket **res)
1372 {
1373         return __sock_create(&init_net, family, type, protocol, res, 1);
1374 }
1375 EXPORT_SYMBOL(sock_create_kern);
1376 
1377 SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol)
1378 {
1379         int retval;
1380         struct socket *sock;
1381         int flags;
1382 
1383         /* 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 
1389         flags = type & ~SOCK_TYPE_MASK;
1390         if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK))
1391                 return -EINVAL;
1392         type &= SOCK_TYPE_MASK;
1393 
1394         if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK))
1395                 flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK;
1396 
1397         retval = sock_create(family, type, protocol, &sock);
1398         if (retval < 0)
1399                 goto out;
1400 
1401         retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK));
1402         if (retval < 0)
1403                 goto out_release;
1404 
1405 out:
1406         /* It may be already another descriptor 8) Not kernel problem. */
1407         return retval;
1408 
1409 out_release:
1410         sock_release(sock);
1411         return retval;
1412 }
1413 
1414 /*
1415  *      Create a pair of connected sockets.
1416  */
1417 
1418 SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol,
1419                 int __user *, usockvec)
1420 {
1421         struct socket *sock1, *sock2;
1422         int fd1, fd2, err;
1423         struct file *newfile1, *newfile2;
1424         int flags;
1425 
1426         flags = type & ~SOCK_TYPE_MASK;
1427         if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK))
1428                 return -EINVAL;
1429         type &= SOCK_TYPE_MASK;
1430 
1431         if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK))
1432                 flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK;
1433 
1434         /*
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);
1448         if (err < 0)
1449                 goto out_release_both;
1450 
1451         fd1 = get_unused_fd_flags(flags);
1452         if (unlikely(fd1 < 0)) {
1453                 err = fd1;
1454                 goto out_release_both;
1455         }
1456 
1457         fd2 = get_unused_fd_flags(flags);
1458         if (unlikely(fd2 < 0)) {
1459                 err = fd2;
1460                 goto out_put_unused_1;
1461         }
1462 
1463         newfile1 = sock_alloc_file(sock1, flags, NULL);
1464         if (unlikely(IS_ERR(newfile1))) {
1465                 err = PTR_ERR(newfile1);
1466                 goto out_put_unused_both;
1467         }
1468 
1469         newfile2 = sock_alloc_file(sock2, flags, NULL);
1470         if (IS_ERR(newfile2)) {
1471                 err = PTR_ERR(newfile2);
1472                 goto out_fput_1;
1473         }
1474 
1475         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 
1483         audit_fd_pair(fd1, fd2);
1484 
1485         fd_install(fd1, newfile1);
1486         fd_install(fd2, newfile2);
1487         /* fd1 and fd2 may be already another descriptors.
1488          * Not kernel problem.
1489          */
1490 
1491         return 0;
1492 
1493 out_fput_both:
1494         fput(newfile2);
1495         fput(newfile1);
1496         put_unused_fd(fd2);
1497         put_unused_fd(fd1);
1498         goto out;
1499 
1500 out_fput_1:
1501         fput(newfile1);
1502         put_unused_fd(fd2);
1503         put_unused_fd(fd1);
1504         sock_release(sock2);
1505         goto out;
1506 
1507 out_put_unused_both:
1508         put_unused_fd(fd2);
1509 out_put_unused_1:
1510         put_unused_fd(fd1);
1511 out_release_both:
1512         sock_release(sock2);
1513 out_release_1:
1514         sock_release(sock1);
1515 out:
1516         return err;
1517 }
1518 
1519 /*
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 
1527 SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen)
1528 {
1529         struct socket *sock;
1530         struct sockaddr_storage address;
1531         int err, fput_needed;
1532 
1533         sock = sockfd_lookup_light(fd, &err, &fput_needed);
1534         if (sock) {
1535                 err = move_addr_to_kernel(umyaddr, addrlen, &address);
1536                 if (err >= 0) {
1537                         err = security_socket_bind(sock,
1538                                                    (struct sockaddr *)&address,
1539                                                    addrlen);
1540                         if (!err)
1541                                 err = sock->ops->bind(sock,
1542                                                       (struct sockaddr *)
1543                                                       &address, addrlen);
1544                 }
1545                 fput_light(sock->file, fput_needed);
1546         }
1547         return err;
1548 }
1549 
1550 /*
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 
1556 SYSCALL_DEFINE2(listen, int, fd, int, backlog)
1557 {
1558         struct socket *sock;
1559         int err, fput_needed;
1560         int somaxconn;
1561 
1562         sock = sockfd_lookup_light(fd, &err, &fput_needed);
1563         if (sock) {
1564                 somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn;
1565                 if ((unsigned int)backlog > somaxconn)
1566                         backlog = somaxconn;
1567 
1568                 err = security_socket_listen(sock, backlog);
1569                 if (!err)
1570                         err = sock->ops->listen(sock, backlog);
1571 
1572                 fput_light(sock->file, fput_needed);
1573         }
1574         return err;
1575 }
1576 
1577 /*
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 
1589 SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr,
1590                 int __user *, upeer_addrlen, int, flags)
1591 {
1592         struct socket *sock, *newsock;
1593         struct file *newfile;
1594         int err, len, newfd, fput_needed;
1595         struct sockaddr_storage address;
1596 
1597         if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK))
1598                 return -EINVAL;
1599 
1600         if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK))
1601                 flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK;
1602 
1603         sock = sockfd_lookup_light(fd, &err, &fput_needed);
1604         if (!sock)
1605                 goto out;
1606 
1607         err = -ENFILE;
1608         newsock = sock_alloc();
1609         if (!newsock)
1610                 goto out_put;
1611 
1612         newsock->type = sock->type;
1613         newsock->ops = sock->ops;
1614 
1615         /*
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 
1621         newfd = get_unused_fd_flags(flags);
1622         if (unlikely(newfd < 0)) {
1623                 err = newfd;
1624                 sock_release(newsock);
1625                 goto out_put;
1626         }
1627         newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name);
1628         if (unlikely(IS_ERR(newfile))) {
1629                 err = PTR_ERR(newfile);
1630                 put_unused_fd(newfd);
1631                 sock_release(newsock);
1632                 goto out_put;
1633         }
1634 
1635         err = security_socket_accept(sock, newsock);
1636         if (err)
1637                 goto out_fd;
1638 
1639         err = sock->ops->accept(sock, newsock, sock->file->f_flags);
1640         if (err < 0)
1641                 goto out_fd;
1642 
1643         if (upeer_sockaddr) {
1644                 if (newsock->ops->getname(newsock, (struct sockaddr *)&address,
1645                                           &len, 2) < 0) {
1646                         err = -ECONNABORTED;
1647                         goto out_fd;
1648                 }
1649                 err = move_addr_to_user(&address,
1650                                         len, upeer_sockaddr, upeer_addrlen);
1651                 if (err < 0)
1652                         goto out_fd;
1653         }
1654 
1655         /* File flags are not inherited via accept() unlike another OSes. */
1656 
1657         fd_install(newfd, newfile);
1658         err = newfd;
1659 
1660 out_put:
1661         fput_light(sock->file, fput_needed);
1662 out:
1663         return err;
1664 out_fd:
1665         fput(newfile);
1666         put_unused_fd(newfd);
1667         goto out_put;
1668 }
1669 
1670 SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr,
1671                 int __user *, upeer_addrlen)
1672 {
1673         return sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0);
1674 }
1675 
1676 /*
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 
1688 SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr,
1689                 int, addrlen)
1690 {
1691         struct socket *sock;
1692         struct sockaddr_storage address;
1693         int err, fput_needed;
1694 
1695         sock = sockfd_lookup_light(fd, &err, &fput_needed);
1696         if (!sock)
1697                 goto out;
1698         err = move_addr_to_kernel(uservaddr, addrlen, &address);
1699         if (err < 0)
1700                 goto out_put;
1701 
1702         err =
1703             security_socket_connect(sock, (struct sockaddr *)&address, addrlen);
1704         if (err)
1705                 goto out_put;
1706 
1707         err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen,
1708                                  sock->file->f_flags);
1709 out_put:
1710         fput_light(sock->file, fput_needed);
1711 out:
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 
1720 SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr,
1721                 int __user *, usockaddr_len)
1722 {
1723         struct socket *sock;
1724         struct sockaddr_storage address;
1725         int len, err, fput_needed;
1726 
1727         sock = sockfd_lookup_light(fd, &err, &fput_needed);
1728         if (!sock)
1729                 goto out;
1730 
1731         err = security_socket_getsockname(sock);
1732         if (err)
1733                 goto out_put;
1734 
1735         err = sock->ops->getname(sock, (struct sockaddr *)&address, &len, 0);
1736         if (err)
1737                 goto out_put;
1738         err = move_addr_to_user(&address, len, usockaddr, usockaddr_len);
1739 
1740 out_put:
1741         fput_light(sock->file, fput_needed);
1742 out:
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 
1751 SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr,
1752                 int __user *, usockaddr_len)
1753 {
1754         struct socket *sock;
1755         struct sockaddr_storage address;
1756         int len, err, fput_needed;
1757 
1758         sock = sockfd_lookup_light(fd, &err, &fput_needed);
1759         if (sock != NULL) {
1760                 err = security_socket_getpeername(sock);
1761                 if (err) {
1762                         fput_light(sock->file, fput_needed);
1763                         return err;
1764                 }
1765 
1766                 err =
1767                     sock->ops->getname(sock, (struct sockaddr *)&address, &len,
1768                                        1);
1769                 if (!err)
1770                         err = move_addr_to_user(&address, len, usockaddr,
1771                                                 usockaddr_len);
1772                 fput_light(sock->file, fput_needed);
1773         }
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 
1783 SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len,
1784                 unsigned int, flags, struct sockaddr __user *, addr,
1785                 int, addr_len)
1786 {
1787         struct socket *sock;
1788         struct sockaddr_storage address;
1789         int err;
1790         struct msghdr msg;
1791         struct iovec iov;
1792         int fput_needed;
1793 
1794         if (len > INT_MAX)
1795                 len = INT_MAX;
1796         sock = sockfd_lookup_light(fd, &err, &fput_needed);
1797         if (!sock)
1798                 goto out;
1799 
1800         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;
1808         if (addr) {
1809                 err = move_addr_to_kernel(addr, addr_len, &address);
1810                 if (err < 0)
1811                         goto out_put;
1812                 msg.msg_name = (struct sockaddr *)&address;
1813                 msg.msg_namelen = addr_len;
1814         }
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 
1820 out_put:
1821         fput_light(sock->file, fput_needed);
1822 out:
1823         return err;
1824 }
1825 
1826 /*
1827  *      Send a datagram down a socket.
1828  */
1829 
1830 SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len,
1831                 unsigned int, flags)
1832 {
1833         return sys_sendto(fd, buff, len, flags, NULL, 0);
1834 }
1835 
1836 /*
1837  *      Receive a frame from the socket and optionally record the address of the
1838  *      sender. We verify the buffers are writable and if needed move the
1839  *      sender address from kernel to user space.
1840  */
1841 
1842 SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size,
1843                 unsigned int, flags, struct sockaddr __user *, addr,
1844                 int __user *, addr_len)
1845 {
1846         struct socket *sock;
1847         struct iovec iov;
1848         struct msghdr msg;
1849         struct sockaddr_storage address;
1850         int err, err2;
1851         int fput_needed;
1852 
1853         if (size > INT_MAX)
1854                 size = INT_MAX;
1855         sock = sockfd_lookup_light(fd, &err, &fput_needed);
1856         if (!sock)
1857                 goto out;
1858 
1859         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;
1865         /* 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;
1869         if (sock->file->f_flags & O_NONBLOCK)
1870                 flags |= MSG_DONTWAIT;
1871         err = sock_recvmsg(sock, &msg, size, flags);
1872 
1873         if (err >= 0 && addr != NULL) {
1874                 err2 = move_addr_to_user(&address,
1875                                          msg.msg_namelen, addr, addr_len);
1876                 if (err2 < 0)
1877                         err = err2;
1878         }
1879 
1880         fput_light(sock->file, fput_needed);
1881 out:
1882         return err;
1883 }
1884 
1885 /*
1886  *      Receive a datagram from a socket.
1887  */
1888 
1889 SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size,
1890                 unsigned int, flags)
1891 {
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 
1900 SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname,
1901                 char __user *, optval, int, optlen)
1902 {
1903         int err, fput_needed;
1904         struct socket *sock;
1905 
1906         if (optlen < 0)
1907                 return -EINVAL;
1908 
1909         sock = sockfd_lookup_light(fd, &err, &fput_needed);
1910         if (sock != NULL) {
1911                 err = security_socket_setsockopt(sock, level, optname);
1912                 if (err)
1913                         goto out_put;
1914 
1915                 if (level == SOL_SOCKET)
1916                         err =
1917                             sock_setsockopt(sock, level, optname, optval,
1918                                             optlen);
1919                 else
1920                         err =
1921                             sock->ops->setsockopt(sock, level, optname, optval,
1922                                                   optlen);
1923 out_put:
1924                 fput_light(sock->file, fput_needed);
1925         }
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 
1934 SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname,
1935                 char __user *, optval, int __user *, optlen)
1936 {
1937         int err, fput_needed;
1938         struct socket *sock;
1939 
1940         sock = sockfd_lookup_light(fd, &err, &fput_needed);
1941         if (sock != NULL) {
1942                 err = security_socket_getsockopt(sock, level, optname);
1943                 if (err)
1944                         goto out_put;
1945 
1946                 if (level == SOL_SOCKET)
1947                         err =
1948                             sock_getsockopt(sock, level, optname, optval,
1949                                             optlen);
1950                 else
1951                         err =
1952                             sock->ops->getsockopt(sock, level, optname, optval,
1953                                                   optlen);
1954 out_put:
1955                 fput_light(sock->file, fput_needed);
1956         }
1957         return err;
1958 }
1959 
1960 /*
1961  *      Shutdown a socket.
1962  */
1963 
1964 SYSCALL_DEFINE2(shutdown, int, fd, int, how)
1965 {
1966         int err, fput_needed;
1967         struct socket *sock;
1968 
1969         sock = sockfd_lookup_light(fd, &err, &fput_needed);
1970         if (sock != NULL) {
1971                 err = security_socket_shutdown(sock, how);
1972                 if (!err)
1973                         err = sock->ops->shutdown(sock, how);
1974                 fput_light(sock->file, fput_needed);
1975         }
1976         return err;
1977 }
1978 
1979 /* A couple of helpful macros for getting the address of the 32/64 bit
1980  * 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 
1986 struct used_address {
1987         struct sockaddr_storage name;
1988         unsigned int name_len;
1989 };
1990 
1991 static 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;
1996 
1997         if (kmsg->msg_name == NULL)
1998                 kmsg->msg_namelen = 0;
1999 
2000         if (kmsg->msg_namelen < 0)
2001                 return -EINVAL;
2002 
2003         if (kmsg->msg_namelen > sizeof(struct sockaddr_storage))
2004                 kmsg->msg_namelen = sizeof(struct sockaddr_storage);
2005         return 0;
2006 }
2007 
2008 static int ___sys_sendmsg(struct socket *sock, struct msghdr __user *msg,
2009                          struct msghdr *msg_sys, unsigned int flags,
2010                          struct used_address *used_address)
2011 {
2012         struct compat_msghdr __user *msg_compat =
2013             (struct compat_msghdr __user *)msg;
2014         struct sockaddr_storage address;
2015         struct iovec iovstack[UIO_FASTIOV], *iov = iovstack;
2016         unsigned char ctl[sizeof(struct cmsghdr) + 20]
2017             __attribute__ ((aligned(sizeof(__kernel_size_t))));
2018         /* 20 is size of ipv6_pktinfo */
2019         unsigned char *ctl_buf = ctl;
2020         int err, ctl_len, total_len;
2021 
2022         err = -EFAULT;
2023         if (MSG_CMSG_COMPAT & flags) {
2024                 if (get_compat_msghdr(msg_sys, msg_compat))
2025                         return -EFAULT;
2026         } else {
2027                 err = copy_msghdr_from_user(msg_sys, msg);
2028                 if (err)
2029                         return err;
2030         }
2031 
2032         if (msg_sys->msg_iovlen > UIO_FASTIOV) {
2033                 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);
2039                 if (!iov)
2040                         goto out;
2041         }
2042 
2043         /* This will also move the address data into kernel space */
2044         if (MSG_CMSG_COMPAT & flags) {
2045                 err = verify_compat_iovec(msg_sys, iov, &address, VERIFY_READ);
2046         } else
2047                 err = verify_iovec(msg_sys, iov, &address, VERIFY_READ);
2048         if (err < 0)
2049                 goto out_freeiov;
2050         total_len = err;
2051 
2052         err = -ENOBUFS;
2053 
2054         if (msg_sys->msg_controllen > INT_MAX)
2055                 goto out_freeiov;
2056         ctl_len = msg_sys->msg_controllen;
2057         if ((MSG_CMSG_COMPAT & flags) && ctl_len) {
2058                 err =
2059                     cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl,
2060                                                      sizeof(ctl));
2061                 if (err)
2062                         goto out_freeiov;
2063                 ctl_buf = msg_sys->msg_control;
2064                 ctl_len = msg_sys->msg_controllen;
2065         } else if (ctl_len) {
2066                 if (ctl_len > sizeof(ctl)) {
2067                         ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL);
2068                         if (ctl_buf == NULL)
2069                                 goto out_freeiov;
2070                 }
2071                 err = -EFAULT;
2072                 /*
2073                  * Careful! Before this, msg_sys->msg_control contains a user pointer.
2074                  * Afterwards, it will be a kernel pointer. Thus the compiler-assisted
2075                  * checking falls down on this.
2076                  */
2077                 if (copy_from_user(ctl_buf,
2078                                    (void __user __force *)msg_sys->msg_control,
2079                                    ctl_len))
2080                         goto out_freectl;
2081                 msg_sys->msg_control = ctl_buf;
2082         }
2083         msg_sys->msg_flags = flags;
2084 
2085         if (sock->file->f_flags & O_NONBLOCK)
2086                 msg_sys->msg_flags |= MSG_DONTWAIT;
2087         /*
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          */
2093         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,
2096                     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;
2107                 if (msg_sys->msg_name)
2108                         memcpy(&used_address->name, msg_sys->msg_name,
2109                                used_address->name_len);
2110         }
2111 
2112 out_freectl:
2113         if (ctl_buf != ctl)
2114                 sock_kfree_s(sock->sk, ctl_buf, ctl_len);
2115 out_freeiov:
2116         if (iov != iovstack)
2117                 kfree(iov);
2118 out:
2119         return err;
2120 }
2121 
2122 /*
2123  *      BSD sendmsg interface
2124  */
2125 
2126 long __sys_sendmsg(int fd, struct msghdr __user *msg, unsigned flags)
2127 {
2128         int fput_needed, err;
2129         struct msghdr msg_sys;
2130         struct socket *sock;
2131 
2132         sock = sockfd_lookup_light(fd, &err, &fput_needed);
2133         if (!sock)
2134                 goto out;
2135 
2136         err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL);
2137 
2138         fput_light(sock->file, fput_needed);
2139 out:
2140         return err;
2141 }
2142 
2143 SYSCALL_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 
2150 /*
2151  *      Linux sendmmsg interface
2152  */
2153 
2154 int __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;
2162         struct used_address used_address;
2163 
2164         if (vlen > UIO_MAXIOV)
2165                 vlen = UIO_MAXIOV;
2166 
2167         datagrams = 0;
2168 
2169         sock = sockfd_lookup_light(fd, &err, &fput_needed);
2170         if (!sock)
2171                 return err;
2172 
2173         used_address.name_len = UINT_MAX;
2174         entry = mmsg;
2175         compat_entry = (struct compat_mmsghdr __user *)mmsg;
2176         err = 0;
2177 
2178         while (datagrams < vlen) {
2179                 if (MSG_CMSG_COMPAT & flags) {
2180                         err = ___sys_sendmsg(sock, (struct msghdr __user *)compat_entry,
2181                                              &msg_sys, flags, &used_address);
2182                         if (err < 0)
2183                                 break;
2184                         err = __put_user(err, &compat_entry->msg_len);
2185                         ++compat_entry;
2186                 } else {
2187                         err = ___sys_sendmsg(sock,
2188                                              (struct msghdr __user *)entry,
2189                                              &msg_sys, flags, &used_address);
2190                         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 
2201         fput_light(sock->file, fput_needed);
2202 
2203         /* We only return an error if no datagrams were able to be sent */
2204         if (datagrams != 0)
2205                 return datagrams;
2206 
2207         return err;
2208 }
2209 
2210 SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg,
2211                 unsigned int, vlen, unsigned int, flags)
2212 {
2213         if (flags & MSG_CMSG_COMPAT)
2214                 return -EINVAL;
2215         return __sys_sendmmsg(fd, mmsg, vlen, flags);
2216 }
2217 
2218 static int ___sys_recvmsg(struct socket *sock, struct msghdr __user *msg,
2219                          struct msghdr *msg_sys, unsigned int flags, int nosec)
2220 {
2221         struct compat_msghdr __user *msg_compat =
2222             (struct compat_msghdr __user *)msg;
2223         struct iovec iovstack[UIO_FASTIOV];
2224         struct iovec *iov = iovstack;
2225         unsigned long cmsg_ptr;
2226         int err, total_len, len;
2227 
2228         /* kernel mode address */
2229         struct sockaddr_storage addr;
2230 
2231         /* user mode address pointers */
2232         struct sockaddr __user *uaddr;
2233         int __user *uaddr_len;
2234 
2235         if (MSG_CMSG_COMPAT & flags) {
2236                 if (get_compat_msghdr(msg_sys, msg_compat))
2237                         return -EFAULT;
2238         } else {
2239                 err = copy_msghdr_from_user(msg_sys, msg);
2240                 if (err)
2241                         return err;
2242         }
2243 
2244         if (msg_sys->msg_iovlen > UIO_FASTIOV) {
2245                 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);
2251                 if (!iov)
2252                         goto out;
2253         }
2254 
2255         /* Save the user-mode address (verify_iovec will change the
2256          * kernel msghdr to use the kernel address space)
2257          */
2258         uaddr = (__force void __user *)msg_sys->msg_name;
2259         uaddr_len = COMPAT_NAMELEN(msg);
2260         if (MSG_CMSG_COMPAT & flags)
2261                 err = verify_compat_iovec(msg_sys, iov, &addr, VERIFY_WRITE);
2262         else
2263                 err = verify_iovec(msg_sys, iov, &addr, VERIFY_WRITE);
2264         if (err < 0)
2265                 goto out_freeiov;
2266         total_len = err;
2267 
2268         cmsg_ptr = (unsigned long)msg_sys->msg_control;
2269         msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT);
2270 
2271         /* We assume all kernel code knows the size of sockaddr_storage */
2272         msg_sys->msg_namelen = 0;
2273 
2274         if (sock->file->f_flags & O_NONBLOCK)
2275                 flags |= MSG_DONTWAIT;
2276         err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys,
2277                                                           total_len, flags);
2278         if (err < 0)
2279                 goto out_freeiov;
2280         len = err;
2281 
2282         if (uaddr != NULL) {
2283                 err = move_addr_to_user(&addr,
2284                                         msg_sys->msg_namelen, uaddr,
2285                                         uaddr_len);
2286                 if (err < 0)
2287                         goto out_freeiov;
2288         }
2289         err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT),
2290                          COMPAT_FLAGS(msg));
2291         if (err)
2292                 goto out_freeiov;
2293         if (MSG_CMSG_COMPAT & flags)
2294                 err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr,
2295                                  &msg_compat->msg_controllen);
2296         else
2297                 err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr,
2298                                  &msg->msg_controllen);
2299         if (err)
2300                 goto out_freeiov;
2301         err = len;
2302 
2303 out_freeiov:
2304         if (iov != iovstack)
2305                 kfree(iov);
2306 out:
2307         return err;
2308 }
2309 
2310 /*
2311  *      BSD recvmsg interface
2312  */
2313 
2314 long __sys_recvmsg(int fd, struct msghdr __user *msg, unsigned flags)
2315 {
2316         int fput_needed, err;
2317         struct msghdr msg_sys;
2318         struct socket *sock;
2319 
2320         sock = sockfd_lookup_light(fd, &err, &fput_needed);
2321         if (!sock)
2322                 goto out;
2323 
2324         err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0);
2325 
2326         fput_light(sock->file, fput_needed);
2327 out:
2328         return err;
2329 }
2330 
2331 SYSCALL_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 
2339 /*
2340  *     Linux recvmmsg interface
2341  */
2342 
2343 int __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;
2349         struct compat_mmsghdr __user *compat_entry;
2350         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;
2369         compat_entry = (struct compat_mmsghdr __user *)mmsg;
2370 
2371         while (datagrams < vlen) {
2372                 /*
2373                  * No need to ask LSM for more than the first datagram.
2374                  */
2375                 if (MSG_CMSG_COMPAT & flags) {
2376                         err = ___sys_recvmsg(sock, (struct msghdr __user *)compat_entry,
2377                                              &msg_sys, flags & ~MSG_WAITFORONE,
2378                                              datagrams);
2379                         if (err < 0)
2380                                 break;
2381                         err = __put_user(err, &compat_entry->msg_len);
2382                         ++compat_entry;
2383                 } else {
2384                         err = ___sys_recvmsg(sock,
2385                                              (struct msghdr __user *)entry,
2386                                              &msg_sys, flags & ~MSG_WAITFORONE,
2387                                              datagrams);
2388                         if (err < 0)
2389                                 break;
2390                         err = put_user(err, &entry->msg_len);
2391                         ++entry;
2392                 }
2393 
2394                 if (err)
2395                         break;
2396                 ++datagrams;
2397 
2398                 /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */
2399                 if (flags & MSG_WAITFORONE)
2400                         flags |= MSG_DONTWAIT;
2401 
2402                 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 
2420 out_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 
2447 SYSCALL_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 
2454         if (flags & MSG_CMSG_COMPAT)
2455                 return -EINVAL;
2456 
2457         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
2473 /* Argument list sizes for sys_socketcall */
2474 #define AL(x) ((x) * sizeof(unsigned long))
2475 static const unsigned char nargs[21] = {
2476         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),
2479         AL(4), AL(5), AL(4)
2480 };
2481 
2482 #undef AL
2483 
2484 /*
2485  *      System call vectors.
2486  *
2487  *      Argument checking cleaned up. Saved 20% in size.
2488  *  This function doesn't need to set the kernel lock because
2489  *  it is set by the callees.
2490  */
2491 
2492 SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args)
2493 {
2494         unsigned long a[AUDITSC_ARGS];
2495         unsigned long a0, a1;
2496         int err;
2497         unsigned int len;
2498 
2499         if (call < 1 || call > SYS_SENDMMSG)
2500                 return -EINVAL;
2501 
2502         len = nargs[call];
2503         if (len > sizeof(a))
2504                 return -EINVAL;
2505 
2506         /* copy_from_user should be SMP safe. */
2507         if (copy_from_user(a, args, len))
2508                 return -EFAULT;
2509 
2510         err = audit_socketcall(nargs[call] / sizeof(unsigned long), a);
2511         if (err)
2512                 return err;
2513 
2514         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:
2531                 err = sys_accept4(a0, (struct sockaddr __user *)a1,
2532                                   (int __user *)a[2], 0);
2533                 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;
2576         case SYS_SENDMMSG:
2577                 err = sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3]);
2578                 break;
2579         case SYS_RECVMSG:
2580                 err = sys_recvmsg(a0, (struct msghdr __user *)a1, a[2]);
2581                 break;
2582         case SYS_RECVMMSG:
2583                 err = sys_recvmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3],
2584                                    (struct timespec __user *)a[4]);
2585                 break;
2586         case SYS_ACCEPT4:
2587                 err = sys_accept4(a0, (struct sockaddr __user *)a1,
2588                                   (int __user *)a[2], a[3]);
2589                 break;
2590         default:
2591                 err = -EINVAL;
2592                 break;
2593         }
2594         return err;
2595 }
2596 
2597 #endif                          /* __ARCH_WANT_SYS_SOCKETCALL */
2598 
2599 /**
2600  *      sock_register - add a socket protocol handler
2601  *      @ops: description of protocol
2602  *
2603  *      This function is called by a protocol handler that wants to
2604  *      advertise its address family, and have it linked into the
2605  *      socket interface. The value ops->family corresponds to the
2606  *      socket system call protocol family.
2607  */
2608 int sock_register(const struct net_proto_family *ops)
2609 {
2610         int err;
2611 
2612         if (ops->family >= NPROTO) {
2613                 pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO);
2614                 return -ENOBUFS;
2615         }
2616 
2617         spin_lock(&net_family_lock);
2618         if (rcu_dereference_protected(net_families[ops->family],
2619                                       lockdep_is_held(&net_family_lock)))
2620                 err = -EEXIST;
2621         else {
2622                 rcu_assign_pointer(net_families[ops->family], ops);
2623                 err = 0;
2624         }
2625         spin_unlock(&net_family_lock);
2626 
2627         pr_info("NET: Registered protocol family %d\n", ops->family);
2628         return err;
2629 }
2630 EXPORT_SYMBOL(sock_register);
2631 
2632 /**
2633  *      sock_unregister - remove a protocol handler
2634  *      @family: protocol family to remove
2635  *
2636  *      This function is called by a protocol handler that wants to
2637  *      remove its address family, and have it unlinked from the
2638  *      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.
2644  */
2645 void sock_unregister(int family)
2646 {
2647         BUG_ON(family < 0 || family >= NPROTO);
2648 
2649         spin_lock(&net_family_lock);
2650         RCU_INIT_POINTER(net_families[family], NULL);
2651         spin_unlock(&net_family_lock);
2652 
2653         synchronize_rcu();
2654 
2655         pr_info("NET: Unregistered protocol family %d\n", family);
2656 }
2657 EXPORT_SYMBOL(sock_unregister);
2658 
2659 static int __init sock_init(void)
2660 {
2661         int err;
2662         /*
2663          *      Initialize the network sysctl infrastructure.
2664          */
2665         err = net_sysctl_init();
2666         if (err)
2667                 goto out;
2668 
2669         /*
2670          *      Initialize skbuff SLAB cache
2671          */
2672         skb_init();
2673 
2674         /*
2675          *      Initialize the protocols module.
2676          */
2677 
2678         init_inodecache();
2679 
2680         err = register_filesystem(&sock_fs_type);
2681         if (err)
2682                 goto out_fs;
2683         sock_mnt = kern_mount(&sock_fs_type);
2684         if (IS_ERR(sock_mnt)) {
2685                 err = PTR_ERR(sock_mnt);
2686                 goto out_mount;
2687         }
2688 
2689         /* The real protocol initialization is performed in later initcalls.
2690          */
2691 
2692 #ifdef CONFIG_NETFILTER
2693         err = netfilter_init();
2694         if (err)
2695                 goto out;
2696 #endif
2697 
2698         ptp_classifier_init();
2699 
2700 out:
2701         return err;
2702 
2703 out_mount:
2704         unregister_filesystem(&sock_fs_type);
2705 out_fs:
2706         goto out;
2707 }
2708 
2709 core_initcall(sock_init);       /* early initcall */
2710 
2711 #ifdef CONFIG_PROC_FS
2712 void socket_seq_show(struct seq_file *seq)
2713 {
2714         int cpu;
2715         int counter = 0;
2716 
2717         for_each_possible_cpu(cpu)
2718             counter += per_cpu(sockets_in_use, cpu);
2719 
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 }
2726 #endif                          /* CONFIG_PROC_FS */
2727 
2728 #ifdef CONFIG_COMPAT
2729 static int do_siocgstamp(struct net *net, struct socket *sock,
2730                          unsigned int cmd, void __user *up)
2731 {
2732         mm_segment_t old_fs = get_fs();
2733         struct timeval ktv;
2734         int err;
2735 
2736         set_fs(KERNEL_DS);
2737         err = sock_do_ioctl(net, sock, cmd, (unsigned long)&ktv);
2738         set_fs(old_fs);
2739         if (!err)
2740                 err = compat_put_timeval(&ktv, up);
2741 
2742         return err;
2743 }
2744 
2745 static int do_siocgstampns(struct net *net, struct socket *sock,
2746                            unsigned int cmd, void __user *up)
2747 {
2748         mm_segment_t old_fs = get_fs();
2749         struct timespec kts;
2750         int err;
2751 
2752         set_fs(KERNEL_DS);
2753         err = sock_do_ioctl(net, sock, cmd, (unsigned long)&kts);
2754         set_fs(old_fs);
2755         if (!err)
2756                 err = compat_put_timespec(&kts, up);
2757 
2758         return err;
2759 }
2760 
2761 static int dev_ifname32(struct net *net, struct compat_ifreq __user *uifr32)
2762 {
2763         struct ifreq __user *uifr;
2764         int err;
2765 
2766         uifr = compat_alloc_user_space(sizeof(struct ifreq));
2767         if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq)))
2768                 return -EFAULT;
2769 
2770         err = dev_ioctl(net, SIOCGIFNAME, uifr);
2771         if (err)
2772                 return err;
2773 
2774         if (copy_in_user(uifr32, uifr, sizeof(struct compat_ifreq)))
2775                 return -EFAULT;
2776 
2777         return 0;
2778 }
2779 
2780 static int dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32)
2781 {
2782         struct compat_ifconf ifc32;
2783         struct ifconf ifc;
2784         struct ifconf __user *uifc;
2785         struct compat_ifreq __user *ifr32;
2786         struct ifreq __user *ifr;
2787         unsigned int i, j;
2788         int err;
2789 
2790         if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf)))
2791                 return -EFAULT;
2792 
2793         memset(&ifc, 0, sizeof(ifc));
2794         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 {
2800                 size_t len = ((ifc32.ifc_len / sizeof(struct compat_ifreq)) + 1) *
2801                         sizeof(struct ifreq);
2802                 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);
2806                 for (i = 0; i < ifc32.ifc_len; i += sizeof(struct compat_ifreq)) {
2807                         if (copy_in_user(ifr, ifr32, sizeof(struct compat_ifreq)))
2808                                 return -EFAULT;
2809                         ifr++;
2810                         ifr32++;
2811                 }
2812         }
2813         if (copy_to_user(uifc, &ifc, sizeof(struct ifconf)))
2814                 return -EFAULT;
2815 
2816         err = dev_ioctl(net, SIOCGIFCONF, uifc);
2817         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;
2826              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)))
2829                         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;
2839                 i = ((i / sizeof(struct ifreq)) * sizeof(struct compat_ifreq));
2840                 ifc32.ifc_len = i;
2841         } else {
2842                 ifc32.ifc_len = i;
2843         }
2844         if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf)))
2845                 return -EFAULT;
2846 
2847         return 0;
2848 }
2849 
2850 static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32)
2851 {
2852         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;
2856         struct ifreq __user *ifr;
2857         u32 rule_cnt = 0, actual_rule_cnt;
2858         u32 ethcmd;
2859         u32 data;
2860         int ret;
2861 
2862         if (get_user(data, &ifr32->ifr_ifru.ifru_data))
2863                 return -EFAULT;
2864 
2865         compat_rxnfc = compat_ptr(data);
2866 
2867         if (get_user(ethcmd, &compat_rxnfc->cmd))
2868                 return -EFAULT;
2869 
2870         /* 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:
2887         case ETHTOOL_SRXCLSRLINS:
2888                 convert_out = true;
2889                 /* fall through */
2890         case ETHTOOL_SRXCLSRLDEL:
2891                 buf_size += sizeof(struct ethtool_rxnfc);
2892                 convert_in = true;
2893                 break;
2894         }
2895 
2896         ifr = compat_alloc_user_space(buf_size);
2897         rxnfc = (void __user *)ifr + ALIGN(sizeof(struct ifreq), 8);
2898 
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) {
2907                 /* We expect there to be holes between fs.m_ext and
2908                  * fs.ring_cookie and at the end of fs, but nowhere else.
2909                  */
2910                 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));
2914                 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,
2921                                  (void __user *)(&rxnfc->fs.m_ext + 1) -
2922                                  (void __user *)rxnfc) ||
2923                     copy_in_user(&rxnfc->fs.ring_cookie,
2924                                  &compat_rxnfc->fs.ring_cookie,
2925                                  (void __user *)(&rxnfc->fs.location + 1) -
2926                                  (void __user *)&rxnfc->fs.ring_cookie) ||
2927                     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,
2938                                  (const void __user *)(&rxnfc->fs.m_ext + 1) -
2939                                  (const void __user *)rxnfc) ||
2940                     copy_in_user(&compat_rxnfc->fs.ring_cookie,
2941                                  &rxnfc->fs.ring_cookie,
2942                                  (const void __user *)(&rxnfc->fs.location + 1) -
2943                                  (const void __user *)&rxnfc->fs.ring_cookie) ||
2944                     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;
2969 }
2970 
2971 static 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 
2977         uifr = compat_alloc_user_space(sizeof(*uifr));
2978         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 
2992 static int bond_ioctl(struct net *net, unsigned int cmd,
2993                          struct compat_ifreq __user *ifr32)
2994 {
2995         struct ifreq kifr;
2996         mm_segment_t old_fs;
2997         int err;
2998 
2999         switch (cmd) {
3000         case SIOCBONDENSLAVE:
3001         case SIOCBONDRELEASE:
3002         case SIOCBONDSETHWADDR:
3003         case SIOCBONDCHANGEACTIVE:
3004                 if (copy_from_user(&kifr, ifr32, sizeof(struct compat_ifreq)))
3005                         return -EFAULT;
3006 
3007                 old_fs = get_fs();
3008                 set_fs(KERNEL_DS);
3009                 err = dev_ioctl(net, cmd,
3010                                 (struct ifreq __user __force *) &kifr);
3011                 set_fs(old_fs);
3012 
3013                 return err;
3014         default:
3015                 return -ENOIOCTLCMD;
3016         }
3017 }
3018 
3019 /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */
3020 static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd,
3021                                  struct compat_ifreq __user *u_ifreq32)
3022 {
3023         struct ifreq __user *u_ifreq64;
3024         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;
3031         if (get_user(data32, &u_ifreq32->ifr_ifru.ifru_data))
3032                 return -EFAULT;
3033         data64 = compat_ptr(data32);
3034 
3035         u_ifreq64 = compat_alloc_user_space(sizeof(*u_ifreq64));
3036 
3037         if (copy_to_user(&u_ifreq64->ifr_ifrn.ifrn_name[0], &tmp_buf[0],
3038                          IFNAMSIZ))
3039                 return -EFAULT;
3040         if (put_user(data64, &u_ifreq64->ifr_ifru.ifru_data))
3041                 return -EFAULT;
3042 
3043         return dev_ioctl(net, cmd, u_ifreq64);
3044 }
3045 
3046 static int dev_ifsioc(struct net *net, struct socket *sock,
3047                          unsigned int cmd, struct compat_ifreq __user *uifr32)
3048 {
3049         struct ifreq __user *uifr;
3050         int err;
3051 
3052         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 
3058         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:
3070                 case SIOCGIFPFLAGS:
3071                 case SIOCGIFTXQLEN:
3072                 case SIOCGMIIPHY:
3073                 case SIOCGMIIREG:
3074                         if (copy_in_user(uifr32, uifr, sizeof(*uifr32)))
3075                                 err = -EFAULT;
3076                         break;
3077                 }
3078         }
3079         return err;
3080 }
3081 
3082 static 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));
3092         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);
3098         if (err)
3099                 return -EFAULT;
3100 
3101         old_fs = get_fs();
3102         set_fs(KERNEL_DS);
3103         err = dev_ioctl(net, cmd, (void  __user __force *)&ifr);
3104         set_fs(old_fs);
3105 
3106         if (cmd == SIOCGIFMAP && !err) {
3107                 err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name));
3108                 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);
3114                 if (err)
3115                         err = -EFAULT;
3116         }
3117         return err;
3118 }
3119 
3120 struct rtentry32 {
3121         u32             rt_pad1;
3122         struct sockaddr rt_dst;         /* target address               */
3123         struct sockaddr rt_gateway;     /* gateway addr (RTF_GATEWAY)   */
3124         struct sockaddr rt_genmask;     /* target network mask (IP)     */
3125         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! */
3132         /* char * */ u32 rt_dev;        /* forcing the device at add    */
3133         u32             rt_mtu;         /* per route MTU/Window         */
3134         u32             rt_window;      /* Window clamping              */
3135         unsigned short  rt_irtt;        /* Initial RTT                  */
3136 };
3137 
3138 struct 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 
3151 static int routing_ioctl(struct net *net, struct socket *sock,
3152                          unsigned int cmd, void __user *argp)
3153 {
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 
3162         if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */
3163                 struct in6_rtmsg32 __user *ur6 = argp;
3164                 ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst),
3165                         3 * sizeof(struct in6_addr));
3166                 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));
3173 
3174                 r = (void *) &r6;
3175         } else { /* ipv4 */
3176                 struct rtentry32 __user *ur4 = argp;
3177                 ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst),
3178                                         3 * sizeof(struct sockaddr));
3179                 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));
3185                 if (rtdev) {
3186                         ret |= copy_from_user(devname, compat_ptr(rtdev), 15);
3187                         r4.rt_dev = (char __user __force *)devname;
3188                         devname[15] = 0;
3189                 } else
3190                         r4.rt_dev = NULL;
3191 
3192                 r = (void *) &r4;
3193         }
3194 
3195         if (ret) {
3196                 ret = -EFAULT;
3197                 goto out;
3198         }
3199 
3200         set_fs(KERNEL_DS);
3201         ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r);
3202         set_fs(old_fs);
3203 
3204 out:
3205         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
3210  * use compatible ioctls
3211  */
3212 static int old_bridge_ioctl(compat_ulong_t __user *argp)
3213 {
3214         compat_ulong_t tmp;
3215 
3216         if (get_user(tmp, argp))
3217                 return -EFAULT;
3218         if (tmp == BRCTL_GET_VERSION)
3219                 return BRCTL_VERSION + 1;
3220         return -EINVAL;
3221 }
3222 
3223 static 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);
3229 
3230         if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15))
3231                 return compat_ifr_data_ioctl(net, cmd, argp);
3232 
3233         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);
3243         case SIOCWANDEV:
3244                 return compat_siocwandev(net, argp);
3245         case SIOCGIFMAP:
3246         case SIOCSIFMAP:
3247                 return compat_sioc_ifmap(net, cmd, argp);
3248         case SIOCBONDENSLAVE:
3249         case SIOCBONDRELEASE:
3250         case SIOCBONDSETHWADDR:
3251         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);
3260         case SIOCBONDSLAVEINFOQUERY:
3261         case SIOCBONDINFOQUERY:
3262         case SIOCSHWTSTAMP:
3263         case SIOCGHWTSTAMP:
3264                 return compat_ifr_data_ioctl(net, cmd, argp);
3265 
3266         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:
3291         case SIOCGIFADDR:
3292         case SIOCSIFADDR:
3293         case SIOCSIFHWBROADCAST:
3294         case SIOCDIFADDR:
3295         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:
3307         case SIOCSIFNAME:
3308         case SIOCGMIIPHY:
3309         case SIOCGMIIREG:
3310         case SIOCSMIIREG:
3311                 return dev_ifsioc(net, sock, cmd, argp);
3312 
3313         case SIOCSARP:
3314         case SIOCGARP:
3315         case SIOCDARP:
3316         case SIOCATMARK:
3317                 return sock_do_ioctl(net, sock, cmd, arg);
3318         }
3319 
3320         return -ENOIOCTLCMD;
3321 }
3322 
3323 static long compat_sock_ioctl(struct file *file, unsigned int cmd,
3324                               unsigned long arg)
3325 {
3326         struct socket *sock = file->private_data;
3327         int ret = -ENOIOCTLCMD;
3328         struct sock *sk;
3329         struct net *net;
3330 
3331         sk = sock->sk;
3332         net = sock_net(sk);
3333 
3334         if (sock->ops->compat_ioctl)
3335                 ret = sock->ops->compat_ioctl(sock, cmd, arg);
3336 
3337         if (ret == -ENOIOCTLCMD &&
3338             (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST))
3339                 ret = compat_wext_handle_ioctl(net, cmd, arg);
3340 
3341         if (ret == -ENOIOCTLCMD)
3342                 ret = compat_sock_ioctl_trans(file, sock, cmd, arg);
3343 
3344         return ret;
3345 }
3346 #endif
3347 
3348 int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen)
3349 {
3350         return sock->ops->bind(sock, addr, addrlen);
3351 }
3352 EXPORT_SYMBOL(kernel_bind);
3353 
3354 int kernel_listen(struct socket *sock, int backlog)
3355 {
3356         return sock->ops->listen(sock, backlog);
3357 }
3358 EXPORT_SYMBOL(kernel_listen);
3359 
3360 int 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);
3373                 *newsock = NULL;
3374                 goto done;
3375         }
3376 
3377         (*newsock)->ops = sock->ops;
3378         __module_get((*newsock)->ops->owner);
3379 
3380 done:
3381         return err;
3382 }
3383 EXPORT_SYMBOL(kernel_accept);
3384 
3385 int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen,
3386                    int flags)
3387 {
3388         return sock->ops->connect(sock, addr, addrlen, flags);
3389 }
3390 EXPORT_SYMBOL(kernel_connect);
3391 
3392 int kernel_getsockname(struct socket *sock, struct sockaddr *addr,
3393                          int *addrlen)
3394 {
3395         return sock->ops->getname(sock, addr, addrlen, 0);
3396 }
3397 EXPORT_SYMBOL(kernel_getsockname);
3398 
3399 int kernel_getpeername(struct socket *sock, struct sockaddr *addr,
3400                          int *addrlen)
3401 {
3402         return sock->ops->getname(sock, addr, addrlen, 1);
3403 }
3404 EXPORT_SYMBOL(kernel_getpeername);
3405 
3406 int kernel_getsockopt(struct socket *sock, int level, int optname,
3407                         char *optval, int *optlen)
3408 {
3409         mm_segment_t oldfs = get_fs();
3410         char __user *uoptval;
3411         int __user *uoptlen;
3412         int err;
3413 
3414         uoptval = (char __user __force *) optval;
3415         uoptlen = (int __user __force *) optlen;
3416 
3417         set_fs(KERNEL_DS);
3418         if (level == SOL_SOCKET)
3419                 err = sock_getsockopt(sock, level, optname, uoptval, uoptlen);
3420         else
3421                 err = sock->ops->getsockopt(sock, level, optname, uoptval,
3422                                             uoptlen);
3423         set_fs(oldfs);
3424         return err;
3425 }
3426 EXPORT_SYMBOL(kernel_getsockopt);
3427 
3428 int kernel_setsockopt(struct socket *sock, int level, int optname,
3429                         char *optval, unsigned int optlen)
3430 {
3431         mm_segment_t oldfs = get_fs();
3432         char __user *uoptval;
3433         int err;
3434 
3435         uoptval = (char __user __force *) optval;
3436 
3437         set_fs(KERNEL_DS);
3438         if (level == SOL_SOCKET)
3439                 err = sock_setsockopt(sock, level, optname, uoptval, optlen);
3440         else
3441                 err = sock->ops->setsockopt(sock, level, optname, uoptval,
3442                                             optlen);
3443         set_fs(oldfs);
3444         return err;
3445 }
3446 EXPORT_SYMBOL(kernel_setsockopt);
3447 
3448 int 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 }
3456 EXPORT_SYMBOL(kernel_sendpage);
3457 
3458 int 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 }
3469 EXPORT_SYMBOL(kernel_sock_ioctl);
3470 
3471 int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how)
3472 {
3473         return sock->ops->shutdown(sock, how);
3474 }
3475 EXPORT_SYMBOL(kernel_sock_shutdown);
3476 

This page was automatically generated by LXR 0.3.1 (source).  •  Linux is a registered trademark of Linus Torvalds  •  Contact us