Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 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> |
Jesper Juhl | 02c30a8 | 2005-05-05 16:16:16 -0700 | [diff] [blame] | 7 | * Ross Biro |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 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/config.h> |
| 62 | #include <linux/mm.h> |
| 63 | #include <linux/smp_lock.h> |
| 64 | #include <linux/socket.h> |
| 65 | #include <linux/file.h> |
| 66 | #include <linux/net.h> |
| 67 | #include <linux/interrupt.h> |
| 68 | #include <linux/netdevice.h> |
| 69 | #include <linux/proc_fs.h> |
| 70 | #include <linux/seq_file.h> |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 71 | #include <linux/mutex.h> |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 72 | #include <linux/wanrouter.h> |
| 73 | #include <linux/if_bridge.h> |
Arnaldo Carvalho de Melo | 2038073 | 2005-08-16 02:18:02 -0300 | [diff] [blame] | 74 | #include <linux/if_frad.h> |
| 75 | #include <linux/if_vlan.h> |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 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/divert.h> |
| 82 | #include <linux/mount.h> |
| 83 | #include <linux/security.h> |
| 84 | #include <linux/syscalls.h> |
| 85 | #include <linux/compat.h> |
| 86 | #include <linux/kmod.h> |
David Woodhouse | 3ec3b2f | 2005-05-17 12:08:48 +0100 | [diff] [blame] | 87 | #include <linux/audit.h> |
Adrian Bunk | d86b5e0 | 2006-01-21 00:46:55 +0100 | [diff] [blame] | 88 | #include <linux/wireless.h> |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 89 | |
| 90 | #include <asm/uaccess.h> |
| 91 | #include <asm/unistd.h> |
| 92 | |
| 93 | #include <net/compat.h> |
| 94 | |
| 95 | #include <net/sock.h> |
| 96 | #include <linux/netfilter.h> |
| 97 | |
| 98 | static int sock_no_open(struct inode *irrelevant, struct file *dontcare); |
| 99 | static ssize_t sock_aio_read(struct kiocb *iocb, char __user *buf, |
| 100 | size_t size, loff_t pos); |
| 101 | static ssize_t sock_aio_write(struct kiocb *iocb, const char __user *buf, |
| 102 | size_t size, loff_t pos); |
| 103 | static int sock_mmap(struct file *file, struct vm_area_struct * vma); |
| 104 | |
| 105 | static int sock_close(struct inode *inode, struct file *file); |
| 106 | static unsigned int sock_poll(struct file *file, |
| 107 | struct poll_table_struct *wait); |
| 108 | static long sock_ioctl(struct file *file, |
| 109 | unsigned int cmd, unsigned long arg); |
Shaun Pereira | 89bbfc9 | 2006-03-21 23:58:08 -0800 | [diff] [blame] | 110 | #ifdef CONFIG_COMPAT |
| 111 | static long compat_sock_ioctl(struct file *file, |
| 112 | unsigned int cmd, unsigned long arg); |
| 113 | #endif |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 114 | static int sock_fasync(int fd, struct file *filp, int on); |
| 115 | static ssize_t sock_readv(struct file *file, const struct iovec *vector, |
| 116 | unsigned long count, loff_t *ppos); |
| 117 | static ssize_t sock_writev(struct file *file, const struct iovec *vector, |
| 118 | unsigned long count, loff_t *ppos); |
| 119 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
| 120 | int offset, size_t size, loff_t *ppos, int more); |
| 121 | |
Jens Axboe | 5274f05 | 2006-03-30 15:15:30 +0200 | [diff] [blame] | 122 | extern ssize_t generic_splice_sendpage(struct inode *inode, struct file *out, |
| 123 | size_t len, unsigned int flags); |
| 124 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 125 | |
| 126 | /* |
| 127 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear |
| 128 | * in the operation structures but are done directly via the socketcall() multiplexor. |
| 129 | */ |
| 130 | |
| 131 | static struct file_operations socket_file_ops = { |
| 132 | .owner = THIS_MODULE, |
| 133 | .llseek = no_llseek, |
| 134 | .aio_read = sock_aio_read, |
| 135 | .aio_write = sock_aio_write, |
| 136 | .poll = sock_poll, |
| 137 | .unlocked_ioctl = sock_ioctl, |
Shaun Pereira | 89bbfc9 | 2006-03-21 23:58:08 -0800 | [diff] [blame] | 138 | #ifdef CONFIG_COMPAT |
| 139 | .compat_ioctl = compat_sock_ioctl, |
| 140 | #endif |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 141 | .mmap = sock_mmap, |
| 142 | .open = sock_no_open, /* special open code to disallow open via /proc */ |
| 143 | .release = sock_close, |
| 144 | .fasync = sock_fasync, |
| 145 | .readv = sock_readv, |
| 146 | .writev = sock_writev, |
Jens Axboe | 5274f05 | 2006-03-30 15:15:30 +0200 | [diff] [blame] | 147 | .sendpage = sock_sendpage, |
| 148 | .splice_write = generic_splice_sendpage, |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 149 | }; |
| 150 | |
| 151 | /* |
| 152 | * The protocol list. Each protocol is registered in here. |
| 153 | */ |
| 154 | |
| 155 | static struct net_proto_family *net_families[NPROTO]; |
| 156 | |
| 157 | #if defined(CONFIG_SMP) || defined(CONFIG_PREEMPT) |
| 158 | static atomic_t net_family_lockct = ATOMIC_INIT(0); |
| 159 | static DEFINE_SPINLOCK(net_family_lock); |
| 160 | |
| 161 | /* The strategy is: modifications net_family vector are short, do not |
| 162 | sleep and veeery rare, but read access should be free of any exclusive |
| 163 | locks. |
| 164 | */ |
| 165 | |
| 166 | static void net_family_write_lock(void) |
| 167 | { |
| 168 | spin_lock(&net_family_lock); |
| 169 | while (atomic_read(&net_family_lockct) != 0) { |
| 170 | spin_unlock(&net_family_lock); |
| 171 | |
| 172 | yield(); |
| 173 | |
| 174 | spin_lock(&net_family_lock); |
| 175 | } |
| 176 | } |
| 177 | |
| 178 | static __inline__ void net_family_write_unlock(void) |
| 179 | { |
| 180 | spin_unlock(&net_family_lock); |
| 181 | } |
| 182 | |
| 183 | static __inline__ void net_family_read_lock(void) |
| 184 | { |
| 185 | atomic_inc(&net_family_lockct); |
| 186 | spin_unlock_wait(&net_family_lock); |
| 187 | } |
| 188 | |
| 189 | static __inline__ void net_family_read_unlock(void) |
| 190 | { |
| 191 | atomic_dec(&net_family_lockct); |
| 192 | } |
| 193 | |
| 194 | #else |
| 195 | #define net_family_write_lock() do { } while(0) |
| 196 | #define net_family_write_unlock() do { } while(0) |
| 197 | #define net_family_read_lock() do { } while(0) |
| 198 | #define net_family_read_unlock() do { } while(0) |
| 199 | #endif |
| 200 | |
| 201 | |
| 202 | /* |
| 203 | * Statistics counters of the socket lists |
| 204 | */ |
| 205 | |
| 206 | static DEFINE_PER_CPU(int, sockets_in_use) = 0; |
| 207 | |
| 208 | /* |
| 209 | * Support routines. Move socket addresses back and forth across the kernel/user |
| 210 | * divide and look after the messy bits. |
| 211 | */ |
| 212 | |
| 213 | #define MAX_SOCK_ADDR 128 /* 108 for Unix domain - |
| 214 | 16 for IP, 16 for IPX, |
| 215 | 24 for IPv6, |
| 216 | about 80 for AX.25 |
| 217 | must be at least one bigger than |
| 218 | the AF_UNIX size (see net/unix/af_unix.c |
| 219 | :unix_mkname()). |
| 220 | */ |
| 221 | |
| 222 | /** |
| 223 | * move_addr_to_kernel - copy a socket address into kernel space |
| 224 | * @uaddr: Address in user space |
| 225 | * @kaddr: Address in kernel space |
| 226 | * @ulen: Length in user space |
| 227 | * |
| 228 | * The address is copied into kernel space. If the provided address is |
| 229 | * too long an error code of -EINVAL is returned. If the copy gives |
| 230 | * invalid addresses -EFAULT is returned. On a success 0 is returned. |
| 231 | */ |
| 232 | |
| 233 | int move_addr_to_kernel(void __user *uaddr, int ulen, void *kaddr) |
| 234 | { |
| 235 | if(ulen<0||ulen>MAX_SOCK_ADDR) |
| 236 | return -EINVAL; |
| 237 | if(ulen==0) |
| 238 | return 0; |
| 239 | if(copy_from_user(kaddr,uaddr,ulen)) |
| 240 | return -EFAULT; |
David Woodhouse | 3ec3b2f | 2005-05-17 12:08:48 +0100 | [diff] [blame] | 241 | return audit_sockaddr(ulen, kaddr); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 242 | } |
| 243 | |
| 244 | /** |
| 245 | * move_addr_to_user - copy an address to user space |
| 246 | * @kaddr: kernel space address |
| 247 | * @klen: length of address in kernel |
| 248 | * @uaddr: user space address |
| 249 | * @ulen: pointer to user length field |
| 250 | * |
| 251 | * The value pointed to by ulen on entry is the buffer length available. |
| 252 | * This is overwritten with the buffer space used. -EINVAL is returned |
| 253 | * if an overlong buffer is specified or a negative buffer size. -EFAULT |
| 254 | * is returned if either the buffer or the length field are not |
| 255 | * accessible. |
| 256 | * After copying the data up to the limit the user specifies, the true |
| 257 | * length of the data is written over the length limit the user |
| 258 | * specified. Zero is returned for a success. |
| 259 | */ |
| 260 | |
| 261 | int move_addr_to_user(void *kaddr, int klen, void __user *uaddr, int __user *ulen) |
| 262 | { |
| 263 | int err; |
| 264 | int len; |
| 265 | |
| 266 | if((err=get_user(len, ulen))) |
| 267 | return err; |
| 268 | if(len>klen) |
| 269 | len=klen; |
| 270 | if(len<0 || len> MAX_SOCK_ADDR) |
| 271 | return -EINVAL; |
| 272 | if(len) |
| 273 | { |
| 274 | if(copy_to_user(uaddr,kaddr,len)) |
| 275 | return -EFAULT; |
| 276 | } |
| 277 | /* |
| 278 | * "fromlen shall refer to the value before truncation.." |
| 279 | * 1003.1g |
| 280 | */ |
| 281 | return __put_user(klen, ulen); |
| 282 | } |
| 283 | |
| 284 | #define SOCKFS_MAGIC 0x534F434B |
| 285 | |
Eric Dumazet | ba89966 | 2005-08-26 12:05:31 -0700 | [diff] [blame] | 286 | static kmem_cache_t * sock_inode_cachep __read_mostly; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 287 | |
| 288 | static struct inode *sock_alloc_inode(struct super_block *sb) |
| 289 | { |
| 290 | struct socket_alloc *ei; |
| 291 | ei = (struct socket_alloc *)kmem_cache_alloc(sock_inode_cachep, SLAB_KERNEL); |
| 292 | if (!ei) |
| 293 | return NULL; |
| 294 | init_waitqueue_head(&ei->socket.wait); |
| 295 | |
| 296 | ei->socket.fasync_list = NULL; |
| 297 | ei->socket.state = SS_UNCONNECTED; |
| 298 | ei->socket.flags = 0; |
| 299 | ei->socket.ops = NULL; |
| 300 | ei->socket.sk = NULL; |
| 301 | ei->socket.file = NULL; |
| 302 | ei->socket.flags = 0; |
| 303 | |
| 304 | return &ei->vfs_inode; |
| 305 | } |
| 306 | |
| 307 | static void sock_destroy_inode(struct inode *inode) |
| 308 | { |
| 309 | kmem_cache_free(sock_inode_cachep, |
| 310 | container_of(inode, struct socket_alloc, vfs_inode)); |
| 311 | } |
| 312 | |
| 313 | static void init_once(void * foo, kmem_cache_t * cachep, unsigned long flags) |
| 314 | { |
| 315 | struct socket_alloc *ei = (struct socket_alloc *) foo; |
| 316 | |
| 317 | if ((flags & (SLAB_CTOR_VERIFY|SLAB_CTOR_CONSTRUCTOR)) == |
| 318 | SLAB_CTOR_CONSTRUCTOR) |
| 319 | inode_init_once(&ei->vfs_inode); |
| 320 | } |
| 321 | |
| 322 | static int init_inodecache(void) |
| 323 | { |
| 324 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", |
| 325 | sizeof(struct socket_alloc), |
Paul Jackson | fffb60f | 2006-03-24 03:16:06 -0800 | [diff] [blame] | 326 | 0, (SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT| |
| 327 | SLAB_MEM_SPREAD), |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 328 | init_once, NULL); |
| 329 | if (sock_inode_cachep == NULL) |
| 330 | return -ENOMEM; |
| 331 | return 0; |
| 332 | } |
| 333 | |
| 334 | static struct super_operations sockfs_ops = { |
| 335 | .alloc_inode = sock_alloc_inode, |
| 336 | .destroy_inode =sock_destroy_inode, |
| 337 | .statfs = simple_statfs, |
| 338 | }; |
| 339 | |
| 340 | static struct super_block *sockfs_get_sb(struct file_system_type *fs_type, |
| 341 | int flags, const char *dev_name, void *data) |
| 342 | { |
| 343 | return get_sb_pseudo(fs_type, "socket:", &sockfs_ops, SOCKFS_MAGIC); |
| 344 | } |
| 345 | |
Eric Dumazet | ba89966 | 2005-08-26 12:05:31 -0700 | [diff] [blame] | 346 | static struct vfsmount *sock_mnt __read_mostly; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 347 | |
| 348 | static struct file_system_type sock_fs_type = { |
| 349 | .name = "sockfs", |
| 350 | .get_sb = sockfs_get_sb, |
| 351 | .kill_sb = kill_anon_super, |
| 352 | }; |
| 353 | static int sockfs_delete_dentry(struct dentry *dentry) |
| 354 | { |
| 355 | return 1; |
| 356 | } |
| 357 | static struct dentry_operations sockfs_dentry_operations = { |
| 358 | .d_delete = sockfs_delete_dentry, |
| 359 | }; |
| 360 | |
| 361 | /* |
| 362 | * Obtains the first available file descriptor and sets it up for use. |
| 363 | * |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 364 | * These functions create file structures and maps them to fd space |
| 365 | * of the current process. On success it returns file descriptor |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 366 | * and file struct implicitly stored in sock->file. |
| 367 | * Note that another thread may close file descriptor before we return |
| 368 | * from this function. We use the fact that now we do not refer |
| 369 | * to socket after mapping. If one day we will need it, this |
| 370 | * function will increment ref. count on file by 1. |
| 371 | * |
| 372 | * In any case returned fd MAY BE not valid! |
| 373 | * This race condition is unavoidable |
| 374 | * with shared fd spaces, we cannot solve it inside kernel, |
| 375 | * but we take care of internal coherence yet. |
| 376 | */ |
| 377 | |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 378 | static int sock_alloc_fd(struct file **filep) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 379 | { |
| 380 | int fd; |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 381 | |
| 382 | fd = get_unused_fd(); |
| 383 | if (likely(fd >= 0)) { |
| 384 | struct file *file = get_empty_filp(); |
| 385 | |
| 386 | *filep = file; |
| 387 | if (unlikely(!file)) { |
| 388 | put_unused_fd(fd); |
| 389 | return -ENFILE; |
| 390 | } |
| 391 | } else |
| 392 | *filep = NULL; |
| 393 | return fd; |
| 394 | } |
| 395 | |
| 396 | static int sock_attach_fd(struct socket *sock, struct file *file) |
| 397 | { |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 398 | struct qstr this; |
| 399 | char name[32]; |
| 400 | |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 401 | this.len = sprintf(name, "[%lu]", SOCK_INODE(sock)->i_ino); |
| 402 | this.name = name; |
| 403 | this.hash = SOCK_INODE(sock)->i_ino; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 404 | |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 405 | file->f_dentry = d_alloc(sock_mnt->mnt_sb->s_root, &this); |
| 406 | if (unlikely(!file->f_dentry)) |
| 407 | return -ENOMEM; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 408 | |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 409 | file->f_dentry->d_op = &sockfs_dentry_operations; |
| 410 | d_add(file->f_dentry, SOCK_INODE(sock)); |
| 411 | file->f_vfsmnt = mntget(sock_mnt); |
| 412 | file->f_mapping = file->f_dentry->d_inode->i_mapping; |
| 413 | |
| 414 | sock->file = file; |
| 415 | file->f_op = SOCK_INODE(sock)->i_fop = &socket_file_ops; |
| 416 | file->f_mode = FMODE_READ | FMODE_WRITE; |
| 417 | file->f_flags = O_RDWR; |
| 418 | file->f_pos = 0; |
| 419 | file->private_data = sock; |
| 420 | |
| 421 | return 0; |
| 422 | } |
| 423 | |
| 424 | int sock_map_fd(struct socket *sock) |
| 425 | { |
| 426 | struct file *newfile; |
| 427 | int fd = sock_alloc_fd(&newfile); |
| 428 | |
| 429 | if (likely(fd >= 0)) { |
| 430 | int err = sock_attach_fd(sock, newfile); |
| 431 | |
| 432 | if (unlikely(err < 0)) { |
| 433 | put_filp(newfile); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 434 | put_unused_fd(fd); |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 435 | return err; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 436 | } |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 437 | fd_install(fd, newfile); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 438 | } |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 439 | return fd; |
| 440 | } |
| 441 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 442 | static struct socket *sock_from_file(struct file *file, int *err) |
| 443 | { |
| 444 | struct inode *inode; |
| 445 | struct socket *sock; |
| 446 | |
| 447 | if (file->f_op == &socket_file_ops) |
| 448 | return file->private_data; /* set in sock_map_fd */ |
| 449 | |
| 450 | inode = file->f_dentry->d_inode; |
| 451 | if (!S_ISSOCK(inode->i_mode)) { |
| 452 | *err = -ENOTSOCK; |
| 453 | return NULL; |
| 454 | } |
| 455 | |
| 456 | sock = SOCKET_I(inode); |
| 457 | if (sock->file != file) { |
| 458 | printk(KERN_ERR "socki_lookup: socket file changed!\n"); |
| 459 | sock->file = file; |
| 460 | } |
| 461 | return sock; |
| 462 | } |
| 463 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 464 | /** |
| 465 | * sockfd_lookup - Go from a file number to its socket slot |
| 466 | * @fd: file handle |
| 467 | * @err: pointer to an error code return |
| 468 | * |
| 469 | * The file handle passed in is locked and the socket it is bound |
| 470 | * too is returned. If an error occurs the err pointer is overwritten |
| 471 | * with a negative errno code and NULL is returned. The function checks |
| 472 | * for both invalid handles and passing a handle which is not a socket. |
| 473 | * |
| 474 | * On a success the socket object pointer is returned. |
| 475 | */ |
| 476 | |
| 477 | struct socket *sockfd_lookup(int fd, int *err) |
| 478 | { |
| 479 | struct file *file; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 480 | struct socket *sock; |
| 481 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 482 | if (!(file = fget(fd))) { |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 483 | *err = -EBADF; |
| 484 | return NULL; |
| 485 | } |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 486 | sock = sock_from_file(file, err); |
| 487 | if (!sock) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 488 | fput(file); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 489 | return sock; |
| 490 | } |
| 491 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 492 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
| 493 | { |
| 494 | struct file *file; |
| 495 | struct socket *sock; |
| 496 | |
| 497 | file = fget_light(fd, fput_needed); |
| 498 | if (file) { |
| 499 | sock = sock_from_file(file, err); |
| 500 | if (sock) |
| 501 | return sock; |
| 502 | fput_light(file, *fput_needed); |
| 503 | } |
| 504 | return NULL; |
| 505 | } |
| 506 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 507 | /** |
| 508 | * sock_alloc - allocate a socket |
| 509 | * |
| 510 | * Allocate a new inode and socket object. The two are bound together |
| 511 | * and initialised. The socket is then returned. If we are out of inodes |
| 512 | * NULL is returned. |
| 513 | */ |
| 514 | |
| 515 | static struct socket *sock_alloc(void) |
| 516 | { |
| 517 | struct inode * inode; |
| 518 | struct socket * sock; |
| 519 | |
| 520 | inode = new_inode(sock_mnt->mnt_sb); |
| 521 | if (!inode) |
| 522 | return NULL; |
| 523 | |
| 524 | sock = SOCKET_I(inode); |
| 525 | |
| 526 | inode->i_mode = S_IFSOCK|S_IRWXUGO; |
| 527 | inode->i_uid = current->fsuid; |
| 528 | inode->i_gid = current->fsgid; |
| 529 | |
| 530 | get_cpu_var(sockets_in_use)++; |
| 531 | put_cpu_var(sockets_in_use); |
| 532 | return sock; |
| 533 | } |
| 534 | |
| 535 | /* |
| 536 | * In theory you can't get an open on this inode, but /proc provides |
| 537 | * a back door. Remember to keep it shut otherwise you'll let the |
| 538 | * creepy crawlies in. |
| 539 | */ |
| 540 | |
| 541 | static int sock_no_open(struct inode *irrelevant, struct file *dontcare) |
| 542 | { |
| 543 | return -ENXIO; |
| 544 | } |
| 545 | |
Arjan van de Ven | 4b6f5d2 | 2006-03-28 01:56:42 -0800 | [diff] [blame] | 546 | const struct file_operations bad_sock_fops = { |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 547 | .owner = THIS_MODULE, |
| 548 | .open = sock_no_open, |
| 549 | }; |
| 550 | |
| 551 | /** |
| 552 | * sock_release - close a socket |
| 553 | * @sock: socket to close |
| 554 | * |
| 555 | * The socket is released from the protocol stack if it has a release |
| 556 | * callback, and the inode is then released if the socket is bound to |
| 557 | * an inode not a file. |
| 558 | */ |
| 559 | |
| 560 | void sock_release(struct socket *sock) |
| 561 | { |
| 562 | if (sock->ops) { |
| 563 | struct module *owner = sock->ops->owner; |
| 564 | |
| 565 | sock->ops->release(sock); |
| 566 | sock->ops = NULL; |
| 567 | module_put(owner); |
| 568 | } |
| 569 | |
| 570 | if (sock->fasync_list) |
| 571 | printk(KERN_ERR "sock_release: fasync list not empty!\n"); |
| 572 | |
| 573 | get_cpu_var(sockets_in_use)--; |
| 574 | put_cpu_var(sockets_in_use); |
| 575 | if (!sock->file) { |
| 576 | iput(SOCK_INODE(sock)); |
| 577 | return; |
| 578 | } |
| 579 | sock->file=NULL; |
| 580 | } |
| 581 | |
| 582 | static inline int __sock_sendmsg(struct kiocb *iocb, struct socket *sock, |
| 583 | struct msghdr *msg, size_t size) |
| 584 | { |
| 585 | struct sock_iocb *si = kiocb_to_siocb(iocb); |
| 586 | int err; |
| 587 | |
| 588 | si->sock = sock; |
| 589 | si->scm = NULL; |
| 590 | si->msg = msg; |
| 591 | si->size = size; |
| 592 | |
| 593 | err = security_socket_sendmsg(sock, msg, size); |
| 594 | if (err) |
| 595 | return err; |
| 596 | |
| 597 | return sock->ops->sendmsg(iocb, sock, msg, size); |
| 598 | } |
| 599 | |
| 600 | int sock_sendmsg(struct socket *sock, struct msghdr *msg, size_t size) |
| 601 | { |
| 602 | struct kiocb iocb; |
| 603 | struct sock_iocb siocb; |
| 604 | int ret; |
| 605 | |
| 606 | init_sync_kiocb(&iocb, NULL); |
| 607 | iocb.private = &siocb; |
| 608 | ret = __sock_sendmsg(&iocb, sock, msg, size); |
| 609 | if (-EIOCBQUEUED == ret) |
| 610 | ret = wait_on_sync_kiocb(&iocb); |
| 611 | return ret; |
| 612 | } |
| 613 | |
| 614 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
| 615 | struct kvec *vec, size_t num, size_t size) |
| 616 | { |
| 617 | mm_segment_t oldfs = get_fs(); |
| 618 | int result; |
| 619 | |
| 620 | set_fs(KERNEL_DS); |
| 621 | /* |
| 622 | * the following is safe, since for compiler definitions of kvec and |
| 623 | * iovec are identical, yielding the same in-core layout and alignment |
| 624 | */ |
| 625 | msg->msg_iov = (struct iovec *)vec, |
| 626 | msg->msg_iovlen = num; |
| 627 | result = sock_sendmsg(sock, msg, size); |
| 628 | set_fs(oldfs); |
| 629 | return result; |
| 630 | } |
| 631 | |
| 632 | static inline int __sock_recvmsg(struct kiocb *iocb, struct socket *sock, |
| 633 | struct msghdr *msg, size_t size, int flags) |
| 634 | { |
| 635 | int err; |
| 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 | si->flags = flags; |
| 643 | |
| 644 | err = security_socket_recvmsg(sock, msg, size, flags); |
| 645 | if (err) |
| 646 | return err; |
| 647 | |
| 648 | return sock->ops->recvmsg(iocb, sock, msg, size, flags); |
| 649 | } |
| 650 | |
| 651 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, |
| 652 | size_t size, int flags) |
| 653 | { |
| 654 | struct kiocb iocb; |
| 655 | struct sock_iocb siocb; |
| 656 | int ret; |
| 657 | |
| 658 | init_sync_kiocb(&iocb, NULL); |
| 659 | iocb.private = &siocb; |
| 660 | ret = __sock_recvmsg(&iocb, sock, msg, size, flags); |
| 661 | if (-EIOCBQUEUED == ret) |
| 662 | ret = wait_on_sync_kiocb(&iocb); |
| 663 | return ret; |
| 664 | } |
| 665 | |
| 666 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
| 667 | struct kvec *vec, size_t num, |
| 668 | size_t size, int flags) |
| 669 | { |
| 670 | mm_segment_t oldfs = get_fs(); |
| 671 | int result; |
| 672 | |
| 673 | set_fs(KERNEL_DS); |
| 674 | /* |
| 675 | * the following is safe, since for compiler definitions of kvec and |
| 676 | * iovec are identical, yielding the same in-core layout and alignment |
| 677 | */ |
| 678 | msg->msg_iov = (struct iovec *)vec, |
| 679 | msg->msg_iovlen = num; |
| 680 | result = sock_recvmsg(sock, msg, size, flags); |
| 681 | set_fs(oldfs); |
| 682 | return result; |
| 683 | } |
| 684 | |
| 685 | static void sock_aio_dtor(struct kiocb *iocb) |
| 686 | { |
| 687 | kfree(iocb->private); |
| 688 | } |
| 689 | |
Arnaldo Carvalho de Melo | 2038073 | 2005-08-16 02:18:02 -0300 | [diff] [blame] | 690 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
| 691 | int offset, size_t size, loff_t *ppos, int more) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 692 | { |
| 693 | struct socket *sock; |
| 694 | int flags; |
| 695 | |
Eric Dumazet | b69aee0 | 2005-09-06 14:42:45 -0700 | [diff] [blame] | 696 | sock = file->private_data; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 697 | |
| 698 | flags = !(file->f_flags & O_NONBLOCK) ? 0 : MSG_DONTWAIT; |
| 699 | if (more) |
| 700 | flags |= MSG_MORE; |
| 701 | |
| 702 | return sock->ops->sendpage(sock, page, offset, size, flags); |
| 703 | } |
| 704 | |
Christoph Hellwig | ce1d4d3 | 2005-12-22 21:08:46 -0800 | [diff] [blame] | 705 | static struct sock_iocb *alloc_sock_iocb(struct kiocb *iocb, |
| 706 | char __user *ubuf, size_t size, struct sock_iocb *siocb) |
| 707 | { |
| 708 | if (!is_sync_kiocb(iocb)) { |
| 709 | siocb = kmalloc(sizeof(*siocb), GFP_KERNEL); |
| 710 | if (!siocb) |
| 711 | return NULL; |
| 712 | iocb->ki_dtor = sock_aio_dtor; |
| 713 | } |
| 714 | |
| 715 | siocb->kiocb = iocb; |
| 716 | siocb->async_iov.iov_base = ubuf; |
| 717 | siocb->async_iov.iov_len = size; |
| 718 | |
| 719 | iocb->private = siocb; |
| 720 | return siocb; |
| 721 | } |
| 722 | |
| 723 | static ssize_t do_sock_read(struct msghdr *msg, struct kiocb *iocb, |
| 724 | struct file *file, struct iovec *iov, unsigned long nr_segs) |
| 725 | { |
| 726 | struct socket *sock = file->private_data; |
| 727 | size_t size = 0; |
| 728 | int i; |
| 729 | |
| 730 | for (i = 0 ; i < nr_segs ; i++) |
| 731 | size += iov[i].iov_len; |
| 732 | |
| 733 | msg->msg_name = NULL; |
| 734 | msg->msg_namelen = 0; |
| 735 | msg->msg_control = NULL; |
| 736 | msg->msg_controllen = 0; |
| 737 | msg->msg_iov = (struct iovec *) iov; |
| 738 | msg->msg_iovlen = nr_segs; |
| 739 | msg->msg_flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
| 740 | |
| 741 | return __sock_recvmsg(iocb, sock, msg, size, msg->msg_flags); |
| 742 | } |
| 743 | |
| 744 | static ssize_t sock_readv(struct file *file, const struct iovec *iov, |
| 745 | unsigned long nr_segs, loff_t *ppos) |
| 746 | { |
| 747 | struct kiocb iocb; |
| 748 | struct sock_iocb siocb; |
| 749 | struct msghdr msg; |
| 750 | int ret; |
| 751 | |
| 752 | init_sync_kiocb(&iocb, NULL); |
| 753 | iocb.private = &siocb; |
| 754 | |
| 755 | ret = do_sock_read(&msg, &iocb, file, (struct iovec *)iov, nr_segs); |
| 756 | if (-EIOCBQUEUED == ret) |
| 757 | ret = wait_on_sync_kiocb(&iocb); |
| 758 | return ret; |
| 759 | } |
| 760 | |
| 761 | static ssize_t sock_aio_read(struct kiocb *iocb, char __user *ubuf, |
| 762 | size_t count, loff_t pos) |
| 763 | { |
| 764 | struct sock_iocb siocb, *x; |
| 765 | |
| 766 | if (pos != 0) |
| 767 | return -ESPIPE; |
| 768 | if (count == 0) /* Match SYS5 behaviour */ |
| 769 | return 0; |
| 770 | |
| 771 | x = alloc_sock_iocb(iocb, ubuf, count, &siocb); |
| 772 | if (!x) |
| 773 | return -ENOMEM; |
| 774 | return do_sock_read(&x->async_msg, iocb, iocb->ki_filp, |
| 775 | &x->async_iov, 1); |
| 776 | } |
| 777 | |
| 778 | static ssize_t do_sock_write(struct msghdr *msg, struct kiocb *iocb, |
| 779 | struct file *file, struct iovec *iov, unsigned long nr_segs) |
| 780 | { |
| 781 | struct socket *sock = file->private_data; |
| 782 | size_t size = 0; |
| 783 | int i; |
| 784 | |
| 785 | for (i = 0 ; i < nr_segs ; i++) |
| 786 | size += iov[i].iov_len; |
| 787 | |
| 788 | msg->msg_name = NULL; |
| 789 | msg->msg_namelen = 0; |
| 790 | msg->msg_control = NULL; |
| 791 | msg->msg_controllen = 0; |
| 792 | msg->msg_iov = (struct iovec *) iov; |
| 793 | msg->msg_iovlen = nr_segs; |
| 794 | msg->msg_flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
| 795 | if (sock->type == SOCK_SEQPACKET) |
| 796 | msg->msg_flags |= MSG_EOR; |
| 797 | |
| 798 | return __sock_sendmsg(iocb, sock, msg, size); |
| 799 | } |
| 800 | |
| 801 | static ssize_t sock_writev(struct file *file, const struct iovec *iov, |
| 802 | unsigned long nr_segs, loff_t *ppos) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 803 | { |
| 804 | struct msghdr msg; |
Christoph Hellwig | ce1d4d3 | 2005-12-22 21:08:46 -0800 | [diff] [blame] | 805 | struct kiocb iocb; |
| 806 | struct sock_iocb siocb; |
| 807 | int ret; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 808 | |
Christoph Hellwig | ce1d4d3 | 2005-12-22 21:08:46 -0800 | [diff] [blame] | 809 | init_sync_kiocb(&iocb, NULL); |
| 810 | iocb.private = &siocb; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 811 | |
Christoph Hellwig | ce1d4d3 | 2005-12-22 21:08:46 -0800 | [diff] [blame] | 812 | ret = do_sock_write(&msg, &iocb, file, (struct iovec *)iov, nr_segs); |
| 813 | if (-EIOCBQUEUED == ret) |
| 814 | ret = wait_on_sync_kiocb(&iocb); |
| 815 | return ret; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 816 | } |
| 817 | |
Christoph Hellwig | ce1d4d3 | 2005-12-22 21:08:46 -0800 | [diff] [blame] | 818 | static ssize_t sock_aio_write(struct kiocb *iocb, const char __user *ubuf, |
| 819 | size_t count, loff_t pos) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 820 | { |
Christoph Hellwig | ce1d4d3 | 2005-12-22 21:08:46 -0800 | [diff] [blame] | 821 | struct sock_iocb siocb, *x; |
| 822 | |
| 823 | if (pos != 0) |
| 824 | return -ESPIPE; |
| 825 | if (count == 0) /* Match SYS5 behaviour */ |
| 826 | return 0; |
| 827 | |
| 828 | x = alloc_sock_iocb(iocb, (void __user *)ubuf, count, &siocb); |
| 829 | if (!x) |
| 830 | return -ENOMEM; |
| 831 | |
| 832 | return do_sock_write(&x->async_msg, iocb, iocb->ki_filp, |
| 833 | &x->async_iov, 1); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 834 | } |
| 835 | |
| 836 | |
| 837 | /* |
| 838 | * Atomic setting of ioctl hooks to avoid race |
| 839 | * with module unload. |
| 840 | */ |
| 841 | |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 842 | static DEFINE_MUTEX(br_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 843 | static int (*br_ioctl_hook)(unsigned int cmd, void __user *arg) = NULL; |
| 844 | |
| 845 | void brioctl_set(int (*hook)(unsigned int, void __user *)) |
| 846 | { |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 847 | mutex_lock(&br_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 848 | br_ioctl_hook = hook; |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 849 | mutex_unlock(&br_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 850 | } |
| 851 | EXPORT_SYMBOL(brioctl_set); |
| 852 | |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 853 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 854 | static int (*vlan_ioctl_hook)(void __user *arg); |
| 855 | |
| 856 | void vlan_ioctl_set(int (*hook)(void __user *)) |
| 857 | { |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 858 | mutex_lock(&vlan_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 859 | vlan_ioctl_hook = hook; |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 860 | mutex_unlock(&vlan_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 861 | } |
| 862 | EXPORT_SYMBOL(vlan_ioctl_set); |
| 863 | |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 864 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 865 | static int (*dlci_ioctl_hook)(unsigned int, void __user *); |
| 866 | |
| 867 | void dlci_ioctl_set(int (*hook)(unsigned int, void __user *)) |
| 868 | { |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 869 | mutex_lock(&dlci_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 870 | dlci_ioctl_hook = hook; |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 871 | mutex_unlock(&dlci_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 872 | } |
| 873 | EXPORT_SYMBOL(dlci_ioctl_set); |
| 874 | |
| 875 | /* |
| 876 | * With an ioctl, arg may well be a user mode pointer, but we don't know |
| 877 | * what to do with it - that's up to the protocol still. |
| 878 | */ |
| 879 | |
| 880 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) |
| 881 | { |
| 882 | struct socket *sock; |
| 883 | void __user *argp = (void __user *)arg; |
| 884 | int pid, err; |
| 885 | |
Eric Dumazet | b69aee0 | 2005-09-06 14:42:45 -0700 | [diff] [blame] | 886 | sock = file->private_data; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 887 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) { |
| 888 | err = dev_ioctl(cmd, argp); |
| 889 | } else |
Adrian Bunk | d86b5e0 | 2006-01-21 00:46:55 +0100 | [diff] [blame] | 890 | #ifdef CONFIG_WIRELESS_EXT |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 891 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
| 892 | err = dev_ioctl(cmd, argp); |
| 893 | } else |
Adrian Bunk | d86b5e0 | 2006-01-21 00:46:55 +0100 | [diff] [blame] | 894 | #endif /* CONFIG_WIRELESS_EXT */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 895 | switch (cmd) { |
| 896 | case FIOSETOWN: |
| 897 | case SIOCSPGRP: |
| 898 | err = -EFAULT; |
| 899 | if (get_user(pid, (int __user *)argp)) |
| 900 | break; |
| 901 | err = f_setown(sock->file, pid, 1); |
| 902 | break; |
| 903 | case FIOGETOWN: |
| 904 | case SIOCGPGRP: |
| 905 | err = put_user(sock->file->f_owner.pid, (int __user *)argp); |
| 906 | break; |
| 907 | case SIOCGIFBR: |
| 908 | case SIOCSIFBR: |
| 909 | case SIOCBRADDBR: |
| 910 | case SIOCBRDELBR: |
| 911 | err = -ENOPKG; |
| 912 | if (!br_ioctl_hook) |
| 913 | request_module("bridge"); |
| 914 | |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 915 | mutex_lock(&br_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 916 | if (br_ioctl_hook) |
| 917 | err = br_ioctl_hook(cmd, argp); |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 918 | mutex_unlock(&br_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 919 | break; |
| 920 | case SIOCGIFVLAN: |
| 921 | case SIOCSIFVLAN: |
| 922 | err = -ENOPKG; |
| 923 | if (!vlan_ioctl_hook) |
| 924 | request_module("8021q"); |
| 925 | |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 926 | mutex_lock(&vlan_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 927 | if (vlan_ioctl_hook) |
| 928 | err = vlan_ioctl_hook(argp); |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 929 | mutex_unlock(&vlan_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 930 | break; |
| 931 | case SIOCGIFDIVERT: |
| 932 | case SIOCSIFDIVERT: |
| 933 | /* Convert this to call through a hook */ |
| 934 | err = divert_ioctl(cmd, argp); |
| 935 | break; |
| 936 | case SIOCADDDLCI: |
| 937 | case SIOCDELDLCI: |
| 938 | err = -ENOPKG; |
| 939 | if (!dlci_ioctl_hook) |
| 940 | request_module("dlci"); |
| 941 | |
| 942 | if (dlci_ioctl_hook) { |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 943 | mutex_lock(&dlci_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 944 | err = dlci_ioctl_hook(cmd, argp); |
Arjan van de Ven | 4a3e2f7 | 2006-03-20 22:33:17 -0800 | [diff] [blame] | 945 | mutex_unlock(&dlci_ioctl_mutex); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 946 | } |
| 947 | break; |
| 948 | default: |
| 949 | err = sock->ops->ioctl(sock, cmd, arg); |
Christoph Hellwig | b5e5fa5 | 2006-01-03 14:18:33 -0800 | [diff] [blame] | 950 | |
| 951 | /* |
| 952 | * If this ioctl is unknown try to hand it down |
| 953 | * to the NIC driver. |
| 954 | */ |
| 955 | if (err == -ENOIOCTLCMD) |
| 956 | err = dev_ioctl(cmd, argp); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 957 | break; |
| 958 | } |
| 959 | return err; |
| 960 | } |
| 961 | |
| 962 | int sock_create_lite(int family, int type, int protocol, struct socket **res) |
| 963 | { |
| 964 | int err; |
| 965 | struct socket *sock = NULL; |
| 966 | |
| 967 | err = security_socket_create(family, type, protocol, 1); |
| 968 | if (err) |
| 969 | goto out; |
| 970 | |
| 971 | sock = sock_alloc(); |
| 972 | if (!sock) { |
| 973 | err = -ENOMEM; |
| 974 | goto out; |
| 975 | } |
| 976 | |
| 977 | security_socket_post_create(sock, family, type, protocol, 1); |
| 978 | sock->type = type; |
| 979 | out: |
| 980 | *res = sock; |
| 981 | return err; |
| 982 | } |
| 983 | |
| 984 | /* No kernel lock held - perfect */ |
| 985 | static unsigned int sock_poll(struct file *file, poll_table * wait) |
| 986 | { |
| 987 | struct socket *sock; |
| 988 | |
| 989 | /* |
| 990 | * We can't return errors to poll, so it's either yes or no. |
| 991 | */ |
Eric Dumazet | b69aee0 | 2005-09-06 14:42:45 -0700 | [diff] [blame] | 992 | sock = file->private_data; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 993 | return sock->ops->poll(file, sock, wait); |
| 994 | } |
| 995 | |
| 996 | static int sock_mmap(struct file * file, struct vm_area_struct * vma) |
| 997 | { |
Eric Dumazet | b69aee0 | 2005-09-06 14:42:45 -0700 | [diff] [blame] | 998 | struct socket *sock = file->private_data; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 999 | |
| 1000 | return sock->ops->mmap(file, sock, vma); |
| 1001 | } |
| 1002 | |
Arnaldo Carvalho de Melo | 2038073 | 2005-08-16 02:18:02 -0300 | [diff] [blame] | 1003 | static int sock_close(struct inode *inode, struct file *filp) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1004 | { |
| 1005 | /* |
| 1006 | * It was possible the inode is NULL we were |
| 1007 | * closing an unfinished socket. |
| 1008 | */ |
| 1009 | |
| 1010 | if (!inode) |
| 1011 | { |
| 1012 | printk(KERN_DEBUG "sock_close: NULL inode\n"); |
| 1013 | return 0; |
| 1014 | } |
| 1015 | sock_fasync(-1, filp, 0); |
| 1016 | sock_release(SOCKET_I(inode)); |
| 1017 | return 0; |
| 1018 | } |
| 1019 | |
| 1020 | /* |
| 1021 | * Update the socket async list |
| 1022 | * |
| 1023 | * Fasync_list locking strategy. |
| 1024 | * |
| 1025 | * 1. fasync_list is modified only under process context socket lock |
| 1026 | * i.e. under semaphore. |
| 1027 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) |
| 1028 | * or under socket lock. |
| 1029 | * 3. fasync_list can be used from softirq context, so that |
| 1030 | * modification under socket lock have to be enhanced with |
| 1031 | * write_lock_bh(&sk->sk_callback_lock). |
| 1032 | * --ANK (990710) |
| 1033 | */ |
| 1034 | |
| 1035 | static int sock_fasync(int fd, struct file *filp, int on) |
| 1036 | { |
| 1037 | struct fasync_struct *fa, *fna=NULL, **prev; |
| 1038 | struct socket *sock; |
| 1039 | struct sock *sk; |
| 1040 | |
| 1041 | if (on) |
| 1042 | { |
Kris Katterjohn | 8b3a700 | 2006-01-11 15:56:43 -0800 | [diff] [blame] | 1043 | fna = kmalloc(sizeof(struct fasync_struct), GFP_KERNEL); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1044 | if(fna==NULL) |
| 1045 | return -ENOMEM; |
| 1046 | } |
| 1047 | |
Eric Dumazet | b69aee0 | 2005-09-06 14:42:45 -0700 | [diff] [blame] | 1048 | sock = filp->private_data; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1049 | |
| 1050 | if ((sk=sock->sk) == NULL) { |
| 1051 | kfree(fna); |
| 1052 | return -EINVAL; |
| 1053 | } |
| 1054 | |
| 1055 | lock_sock(sk); |
| 1056 | |
| 1057 | prev=&(sock->fasync_list); |
| 1058 | |
| 1059 | for (fa=*prev; fa!=NULL; prev=&fa->fa_next,fa=*prev) |
| 1060 | if (fa->fa_file==filp) |
| 1061 | break; |
| 1062 | |
| 1063 | if(on) |
| 1064 | { |
| 1065 | if(fa!=NULL) |
| 1066 | { |
| 1067 | write_lock_bh(&sk->sk_callback_lock); |
| 1068 | fa->fa_fd=fd; |
| 1069 | write_unlock_bh(&sk->sk_callback_lock); |
| 1070 | |
| 1071 | kfree(fna); |
| 1072 | goto out; |
| 1073 | } |
| 1074 | fna->fa_file=filp; |
| 1075 | fna->fa_fd=fd; |
| 1076 | fna->magic=FASYNC_MAGIC; |
| 1077 | fna->fa_next=sock->fasync_list; |
| 1078 | write_lock_bh(&sk->sk_callback_lock); |
| 1079 | sock->fasync_list=fna; |
| 1080 | write_unlock_bh(&sk->sk_callback_lock); |
| 1081 | } |
| 1082 | else |
| 1083 | { |
| 1084 | if (fa!=NULL) |
| 1085 | { |
| 1086 | write_lock_bh(&sk->sk_callback_lock); |
| 1087 | *prev=fa->fa_next; |
| 1088 | write_unlock_bh(&sk->sk_callback_lock); |
| 1089 | kfree(fa); |
| 1090 | } |
| 1091 | } |
| 1092 | |
| 1093 | out: |
| 1094 | release_sock(sock->sk); |
| 1095 | return 0; |
| 1096 | } |
| 1097 | |
| 1098 | /* This function may be called only under socket lock or callback_lock */ |
| 1099 | |
| 1100 | int sock_wake_async(struct socket *sock, int how, int band) |
| 1101 | { |
| 1102 | if (!sock || !sock->fasync_list) |
| 1103 | return -1; |
| 1104 | switch (how) |
| 1105 | { |
| 1106 | case 1: |
| 1107 | |
| 1108 | if (test_bit(SOCK_ASYNC_WAITDATA, &sock->flags)) |
| 1109 | break; |
| 1110 | goto call_kill; |
| 1111 | case 2: |
| 1112 | if (!test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags)) |
| 1113 | break; |
| 1114 | /* fall through */ |
| 1115 | case 0: |
| 1116 | call_kill: |
| 1117 | __kill_fasync(sock->fasync_list, SIGIO, band); |
| 1118 | break; |
| 1119 | case 3: |
| 1120 | __kill_fasync(sock->fasync_list, SIGURG, band); |
| 1121 | } |
| 1122 | return 0; |
| 1123 | } |
| 1124 | |
| 1125 | static int __sock_create(int family, int type, int protocol, struct socket **res, int kern) |
| 1126 | { |
| 1127 | int err; |
| 1128 | struct socket *sock; |
| 1129 | |
| 1130 | /* |
| 1131 | * Check protocol is in range |
| 1132 | */ |
| 1133 | if (family < 0 || family >= NPROTO) |
| 1134 | return -EAFNOSUPPORT; |
| 1135 | if (type < 0 || type >= SOCK_MAX) |
| 1136 | return -EINVAL; |
| 1137 | |
| 1138 | /* Compatibility. |
| 1139 | |
| 1140 | This uglymoron is moved from INET layer to here to avoid |
| 1141 | deadlock in module load. |
| 1142 | */ |
| 1143 | if (family == PF_INET && type == SOCK_PACKET) { |
| 1144 | static int warned; |
| 1145 | if (!warned) { |
| 1146 | warned = 1; |
| 1147 | printk(KERN_INFO "%s uses obsolete (PF_INET,SOCK_PACKET)\n", current->comm); |
| 1148 | } |
| 1149 | family = PF_PACKET; |
| 1150 | } |
| 1151 | |
| 1152 | err = security_socket_create(family, type, protocol, kern); |
| 1153 | if (err) |
| 1154 | return err; |
| 1155 | |
| 1156 | #if defined(CONFIG_KMOD) |
| 1157 | /* Attempt to load a protocol module if the find failed. |
| 1158 | * |
| 1159 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user |
| 1160 | * requested real, full-featured networking support upon configuration. |
| 1161 | * Otherwise module support will break! |
| 1162 | */ |
| 1163 | if (net_families[family]==NULL) |
| 1164 | { |
| 1165 | request_module("net-pf-%d",family); |
| 1166 | } |
| 1167 | #endif |
| 1168 | |
| 1169 | net_family_read_lock(); |
| 1170 | if (net_families[family] == NULL) { |
| 1171 | err = -EAFNOSUPPORT; |
| 1172 | goto out; |
| 1173 | } |
| 1174 | |
| 1175 | /* |
| 1176 | * Allocate the socket and allow the family to set things up. if |
| 1177 | * the protocol is 0, the family is instructed to select an appropriate |
| 1178 | * default. |
| 1179 | */ |
| 1180 | |
| 1181 | if (!(sock = sock_alloc())) { |
| 1182 | printk(KERN_WARNING "socket: no more sockets\n"); |
| 1183 | err = -ENFILE; /* Not exactly a match, but its the |
| 1184 | closest posix thing */ |
| 1185 | goto out; |
| 1186 | } |
| 1187 | |
| 1188 | sock->type = type; |
| 1189 | |
| 1190 | /* |
| 1191 | * We will call the ->create function, that possibly is in a loadable |
| 1192 | * module, so we have to bump that loadable module refcnt first. |
| 1193 | */ |
| 1194 | err = -EAFNOSUPPORT; |
| 1195 | if (!try_module_get(net_families[family]->owner)) |
| 1196 | goto out_release; |
| 1197 | |
Frank Filz | a79af59 | 2005-09-27 15:23:38 -0700 | [diff] [blame] | 1198 | if ((err = net_families[family]->create(sock, protocol)) < 0) { |
| 1199 | sock->ops = NULL; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1200 | goto out_module_put; |
Frank Filz | a79af59 | 2005-09-27 15:23:38 -0700 | [diff] [blame] | 1201 | } |
| 1202 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1203 | /* |
| 1204 | * Now to bump the refcnt of the [loadable] module that owns this |
| 1205 | * socket at sock_release time we decrement its refcnt. |
| 1206 | */ |
| 1207 | if (!try_module_get(sock->ops->owner)) { |
| 1208 | sock->ops = NULL; |
| 1209 | goto out_module_put; |
| 1210 | } |
| 1211 | /* |
| 1212 | * Now that we're done with the ->create function, the [loadable] |
| 1213 | * module can have its refcnt decremented |
| 1214 | */ |
| 1215 | module_put(net_families[family]->owner); |
| 1216 | *res = sock; |
| 1217 | security_socket_post_create(sock, family, type, protocol, kern); |
| 1218 | |
| 1219 | out: |
| 1220 | net_family_read_unlock(); |
| 1221 | return err; |
| 1222 | out_module_put: |
| 1223 | module_put(net_families[family]->owner); |
| 1224 | out_release: |
| 1225 | sock_release(sock); |
| 1226 | goto out; |
| 1227 | } |
| 1228 | |
| 1229 | int sock_create(int family, int type, int protocol, struct socket **res) |
| 1230 | { |
| 1231 | return __sock_create(family, type, protocol, res, 0); |
| 1232 | } |
| 1233 | |
| 1234 | int sock_create_kern(int family, int type, int protocol, struct socket **res) |
| 1235 | { |
| 1236 | return __sock_create(family, type, protocol, res, 1); |
| 1237 | } |
| 1238 | |
| 1239 | asmlinkage long sys_socket(int family, int type, int protocol) |
| 1240 | { |
| 1241 | int retval; |
| 1242 | struct socket *sock; |
| 1243 | |
| 1244 | retval = sock_create(family, type, protocol, &sock); |
| 1245 | if (retval < 0) |
| 1246 | goto out; |
| 1247 | |
| 1248 | retval = sock_map_fd(sock); |
| 1249 | if (retval < 0) |
| 1250 | goto out_release; |
| 1251 | |
| 1252 | out: |
| 1253 | /* It may be already another descriptor 8) Not kernel problem. */ |
| 1254 | return retval; |
| 1255 | |
| 1256 | out_release: |
| 1257 | sock_release(sock); |
| 1258 | return retval; |
| 1259 | } |
| 1260 | |
| 1261 | /* |
| 1262 | * Create a pair of connected sockets. |
| 1263 | */ |
| 1264 | |
| 1265 | asmlinkage long sys_socketpair(int family, int type, int protocol, int __user *usockvec) |
| 1266 | { |
| 1267 | struct socket *sock1, *sock2; |
| 1268 | int fd1, fd2, err; |
| 1269 | |
| 1270 | /* |
| 1271 | * Obtain the first socket and check if the underlying protocol |
| 1272 | * supports the socketpair call. |
| 1273 | */ |
| 1274 | |
| 1275 | err = sock_create(family, type, protocol, &sock1); |
| 1276 | if (err < 0) |
| 1277 | goto out; |
| 1278 | |
| 1279 | err = sock_create(family, type, protocol, &sock2); |
| 1280 | if (err < 0) |
| 1281 | goto out_release_1; |
| 1282 | |
| 1283 | err = sock1->ops->socketpair(sock1, sock2); |
| 1284 | if (err < 0) |
| 1285 | goto out_release_both; |
| 1286 | |
| 1287 | fd1 = fd2 = -1; |
| 1288 | |
| 1289 | err = sock_map_fd(sock1); |
| 1290 | if (err < 0) |
| 1291 | goto out_release_both; |
| 1292 | fd1 = err; |
| 1293 | |
| 1294 | err = sock_map_fd(sock2); |
| 1295 | if (err < 0) |
| 1296 | goto out_close_1; |
| 1297 | fd2 = err; |
| 1298 | |
| 1299 | /* fd1 and fd2 may be already another descriptors. |
| 1300 | * Not kernel problem. |
| 1301 | */ |
| 1302 | |
| 1303 | err = put_user(fd1, &usockvec[0]); |
| 1304 | if (!err) |
| 1305 | err = put_user(fd2, &usockvec[1]); |
| 1306 | if (!err) |
| 1307 | return 0; |
| 1308 | |
| 1309 | sys_close(fd2); |
| 1310 | sys_close(fd1); |
| 1311 | return err; |
| 1312 | |
| 1313 | out_close_1: |
| 1314 | sock_release(sock2); |
| 1315 | sys_close(fd1); |
| 1316 | return err; |
| 1317 | |
| 1318 | out_release_both: |
| 1319 | sock_release(sock2); |
| 1320 | out_release_1: |
| 1321 | sock_release(sock1); |
| 1322 | out: |
| 1323 | return err; |
| 1324 | } |
| 1325 | |
| 1326 | |
| 1327 | /* |
| 1328 | * Bind a name to a socket. Nothing much to do here since it's |
| 1329 | * the protocol's responsibility to handle the local address. |
| 1330 | * |
| 1331 | * We move the socket address to kernel space before we call |
| 1332 | * the protocol layer (having also checked the address is ok). |
| 1333 | */ |
| 1334 | |
| 1335 | asmlinkage long sys_bind(int fd, struct sockaddr __user *umyaddr, int addrlen) |
| 1336 | { |
| 1337 | struct socket *sock; |
| 1338 | char address[MAX_SOCK_ADDR]; |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1339 | int err, fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1340 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1341 | if((sock = sockfd_lookup_light(fd, &err, &fput_needed))!=NULL) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1342 | { |
| 1343 | if((err=move_addr_to_kernel(umyaddr,addrlen,address))>=0) { |
| 1344 | err = security_socket_bind(sock, (struct sockaddr *)address, addrlen); |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1345 | if (!err) |
| 1346 | err = sock->ops->bind(sock, |
| 1347 | (struct sockaddr *)address, addrlen); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1348 | } |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1349 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1350 | } |
| 1351 | return err; |
| 1352 | } |
| 1353 | |
| 1354 | |
| 1355 | /* |
| 1356 | * Perform a listen. Basically, we allow the protocol to do anything |
| 1357 | * necessary for a listen, and if that works, we mark the socket as |
| 1358 | * ready for listening. |
| 1359 | */ |
| 1360 | |
| 1361 | int sysctl_somaxconn = SOMAXCONN; |
| 1362 | |
| 1363 | asmlinkage long sys_listen(int fd, int backlog) |
| 1364 | { |
| 1365 | struct socket *sock; |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1366 | int err, fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1367 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1368 | if ((sock = sockfd_lookup_light(fd, &err, &fput_needed)) != NULL) { |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1369 | if ((unsigned) backlog > sysctl_somaxconn) |
| 1370 | backlog = sysctl_somaxconn; |
| 1371 | |
| 1372 | err = security_socket_listen(sock, backlog); |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1373 | if (!err) |
| 1374 | err = sock->ops->listen(sock, backlog); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1375 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1376 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1377 | } |
| 1378 | return err; |
| 1379 | } |
| 1380 | |
| 1381 | |
| 1382 | /* |
| 1383 | * For accept, we attempt to create a new socket, set up the link |
| 1384 | * with the client, wake up the client, then return the new |
| 1385 | * connected fd. We collect the address of the connector in kernel |
| 1386 | * space and move it to user at the very end. This is unclean because |
| 1387 | * we open the socket then return an error. |
| 1388 | * |
| 1389 | * 1003.1g adds the ability to recvmsg() to query connection pending |
| 1390 | * status to recvmsg. We need to add that support in a way thats |
| 1391 | * clean when we restucture accept also. |
| 1392 | */ |
| 1393 | |
| 1394 | asmlinkage long sys_accept(int fd, struct sockaddr __user *upeer_sockaddr, int __user *upeer_addrlen) |
| 1395 | { |
| 1396 | struct socket *sock, *newsock; |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 1397 | struct file *newfile; |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1398 | int err, len, newfd, fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1399 | char address[MAX_SOCK_ADDR]; |
| 1400 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1401 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1402 | if (!sock) |
| 1403 | goto out; |
| 1404 | |
| 1405 | err = -ENFILE; |
| 1406 | if (!(newsock = sock_alloc())) |
| 1407 | goto out_put; |
| 1408 | |
| 1409 | newsock->type = sock->type; |
| 1410 | newsock->ops = sock->ops; |
| 1411 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1412 | /* |
| 1413 | * We don't need try_module_get here, as the listening socket (sock) |
| 1414 | * has the protocol module (sock->ops->owner) held. |
| 1415 | */ |
| 1416 | __module_get(newsock->ops->owner); |
| 1417 | |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 1418 | newfd = sock_alloc_fd(&newfile); |
| 1419 | if (unlikely(newfd < 0)) { |
| 1420 | err = newfd; |
David S. Miller | 9a1875e | 2006-04-01 12:48:36 -0800 | [diff] [blame] | 1421 | sock_release(newsock); |
| 1422 | goto out_put; |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 1423 | } |
| 1424 | |
| 1425 | err = sock_attach_fd(newsock, newfile); |
| 1426 | if (err < 0) |
| 1427 | goto out_fd; |
| 1428 | |
Frank Filz | a79af59 | 2005-09-27 15:23:38 -0700 | [diff] [blame] | 1429 | err = security_socket_accept(sock, newsock); |
| 1430 | if (err) |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 1431 | goto out_fd; |
Frank Filz | a79af59 | 2005-09-27 15:23:38 -0700 | [diff] [blame] | 1432 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1433 | err = sock->ops->accept(sock, newsock, sock->file->f_flags); |
| 1434 | if (err < 0) |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 1435 | goto out_fd; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1436 | |
| 1437 | if (upeer_sockaddr) { |
| 1438 | if(newsock->ops->getname(newsock, (struct sockaddr *)address, &len, 2)<0) { |
| 1439 | err = -ECONNABORTED; |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 1440 | goto out_fd; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1441 | } |
| 1442 | err = move_addr_to_user(address, len, upeer_sockaddr, upeer_addrlen); |
| 1443 | if (err < 0) |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 1444 | goto out_fd; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1445 | } |
| 1446 | |
| 1447 | /* File flags are not inherited via accept() unlike another OSes. */ |
| 1448 | |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 1449 | fd_install(newfd, newfile); |
| 1450 | err = newfd; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1451 | |
| 1452 | security_socket_post_accept(sock, newsock); |
| 1453 | |
| 1454 | out_put: |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1455 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1456 | out: |
| 1457 | return err; |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 1458 | out_fd: |
David S. Miller | 9606a21 | 2006-04-01 01:00:14 -0800 | [diff] [blame] | 1459 | fput(newfile); |
David S. Miller | 39d8c1b | 2006-03-20 17:13:49 -0800 | [diff] [blame] | 1460 | put_unused_fd(newfd); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1461 | goto out_put; |
| 1462 | } |
| 1463 | |
| 1464 | |
| 1465 | /* |
| 1466 | * Attempt to connect to a socket with the server address. The address |
| 1467 | * is in user space so we verify it is OK and move it to kernel space. |
| 1468 | * |
| 1469 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to |
| 1470 | * break bindings |
| 1471 | * |
| 1472 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and |
| 1473 | * other SEQPACKET protocols that take time to connect() as it doesn't |
| 1474 | * include the -EINPROGRESS status for such sockets. |
| 1475 | */ |
| 1476 | |
| 1477 | asmlinkage long sys_connect(int fd, struct sockaddr __user *uservaddr, int addrlen) |
| 1478 | { |
| 1479 | struct socket *sock; |
| 1480 | char address[MAX_SOCK_ADDR]; |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1481 | int err, fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1482 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1483 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1484 | if (!sock) |
| 1485 | goto out; |
| 1486 | err = move_addr_to_kernel(uservaddr, addrlen, address); |
| 1487 | if (err < 0) |
| 1488 | goto out_put; |
| 1489 | |
| 1490 | err = security_socket_connect(sock, (struct sockaddr *)address, addrlen); |
| 1491 | if (err) |
| 1492 | goto out_put; |
| 1493 | |
| 1494 | err = sock->ops->connect(sock, (struct sockaddr *) address, addrlen, |
| 1495 | sock->file->f_flags); |
| 1496 | out_put: |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1497 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1498 | out: |
| 1499 | return err; |
| 1500 | } |
| 1501 | |
| 1502 | /* |
| 1503 | * Get the local address ('name') of a socket object. Move the obtained |
| 1504 | * name to user space. |
| 1505 | */ |
| 1506 | |
| 1507 | asmlinkage long sys_getsockname(int fd, struct sockaddr __user *usockaddr, int __user *usockaddr_len) |
| 1508 | { |
| 1509 | struct socket *sock; |
| 1510 | char address[MAX_SOCK_ADDR]; |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1511 | int len, err, fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1512 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1513 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1514 | if (!sock) |
| 1515 | goto out; |
| 1516 | |
| 1517 | err = security_socket_getsockname(sock); |
| 1518 | if (err) |
| 1519 | goto out_put; |
| 1520 | |
| 1521 | err = sock->ops->getname(sock, (struct sockaddr *)address, &len, 0); |
| 1522 | if (err) |
| 1523 | goto out_put; |
| 1524 | err = move_addr_to_user(address, len, usockaddr, usockaddr_len); |
| 1525 | |
| 1526 | out_put: |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1527 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1528 | out: |
| 1529 | return err; |
| 1530 | } |
| 1531 | |
| 1532 | /* |
| 1533 | * Get the remote address ('name') of a socket object. Move the obtained |
| 1534 | * name to user space. |
| 1535 | */ |
| 1536 | |
| 1537 | asmlinkage long sys_getpeername(int fd, struct sockaddr __user *usockaddr, int __user *usockaddr_len) |
| 1538 | { |
| 1539 | struct socket *sock; |
| 1540 | char address[MAX_SOCK_ADDR]; |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1541 | int len, err, fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1542 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1543 | if ((sock = sockfd_lookup_light(fd, &err, &fput_needed)) != NULL) { |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1544 | err = security_socket_getpeername(sock); |
| 1545 | if (err) { |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1546 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1547 | return err; |
| 1548 | } |
| 1549 | |
| 1550 | err = sock->ops->getname(sock, (struct sockaddr *)address, &len, 1); |
| 1551 | if (!err) |
| 1552 | err=move_addr_to_user(address,len, usockaddr, usockaddr_len); |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1553 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1554 | } |
| 1555 | return err; |
| 1556 | } |
| 1557 | |
| 1558 | /* |
| 1559 | * Send a datagram to a given address. We move the address into kernel |
| 1560 | * space and check the user space data area is readable before invoking |
| 1561 | * the protocol. |
| 1562 | */ |
| 1563 | |
| 1564 | asmlinkage long sys_sendto(int fd, void __user * buff, size_t len, unsigned flags, |
| 1565 | struct sockaddr __user *addr, int addr_len) |
| 1566 | { |
| 1567 | struct socket *sock; |
| 1568 | char address[MAX_SOCK_ADDR]; |
| 1569 | int err; |
| 1570 | struct msghdr msg; |
| 1571 | struct iovec iov; |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1572 | int fput_needed; |
| 1573 | struct file *sock_file; |
| 1574 | |
| 1575 | sock_file = fget_light(fd, &fput_needed); |
| 1576 | if (!sock_file) |
| 1577 | return -EBADF; |
| 1578 | |
| 1579 | sock = sock_from_file(sock_file, &err); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1580 | if (!sock) |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1581 | goto out_put; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1582 | iov.iov_base=buff; |
| 1583 | iov.iov_len=len; |
| 1584 | msg.msg_name=NULL; |
| 1585 | msg.msg_iov=&iov; |
| 1586 | msg.msg_iovlen=1; |
| 1587 | msg.msg_control=NULL; |
| 1588 | msg.msg_controllen=0; |
| 1589 | msg.msg_namelen=0; |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1590 | if (addr) { |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1591 | err = move_addr_to_kernel(addr, addr_len, address); |
| 1592 | if (err < 0) |
| 1593 | goto out_put; |
| 1594 | msg.msg_name=address; |
| 1595 | msg.msg_namelen=addr_len; |
| 1596 | } |
| 1597 | if (sock->file->f_flags & O_NONBLOCK) |
| 1598 | flags |= MSG_DONTWAIT; |
| 1599 | msg.msg_flags = flags; |
| 1600 | err = sock_sendmsg(sock, &msg, len); |
| 1601 | |
| 1602 | out_put: |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1603 | fput_light(sock_file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1604 | return err; |
| 1605 | } |
| 1606 | |
| 1607 | /* |
| 1608 | * Send a datagram down a socket. |
| 1609 | */ |
| 1610 | |
| 1611 | asmlinkage long sys_send(int fd, void __user * buff, size_t len, unsigned flags) |
| 1612 | { |
| 1613 | return sys_sendto(fd, buff, len, flags, NULL, 0); |
| 1614 | } |
| 1615 | |
| 1616 | /* |
| 1617 | * Receive a frame from the socket and optionally record the address of the |
| 1618 | * sender. We verify the buffers are writable and if needed move the |
| 1619 | * sender address from kernel to user space. |
| 1620 | */ |
| 1621 | |
| 1622 | asmlinkage long sys_recvfrom(int fd, void __user * ubuf, size_t size, unsigned flags, |
| 1623 | struct sockaddr __user *addr, int __user *addr_len) |
| 1624 | { |
| 1625 | struct socket *sock; |
| 1626 | struct iovec iov; |
| 1627 | struct msghdr msg; |
| 1628 | char address[MAX_SOCK_ADDR]; |
| 1629 | int err,err2; |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1630 | struct file *sock_file; |
| 1631 | int fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1632 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1633 | sock_file = fget_light(fd, &fput_needed); |
| 1634 | if (!sock_file) |
| 1635 | return -EBADF; |
| 1636 | |
| 1637 | sock = sock_from_file(sock_file, &err); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1638 | if (!sock) |
| 1639 | goto out; |
| 1640 | |
| 1641 | msg.msg_control=NULL; |
| 1642 | msg.msg_controllen=0; |
| 1643 | msg.msg_iovlen=1; |
| 1644 | msg.msg_iov=&iov; |
| 1645 | iov.iov_len=size; |
| 1646 | iov.iov_base=ubuf; |
| 1647 | msg.msg_name=address; |
| 1648 | msg.msg_namelen=MAX_SOCK_ADDR; |
| 1649 | if (sock->file->f_flags & O_NONBLOCK) |
| 1650 | flags |= MSG_DONTWAIT; |
| 1651 | err=sock_recvmsg(sock, &msg, size, flags); |
| 1652 | |
| 1653 | if(err >= 0 && addr != NULL) |
| 1654 | { |
| 1655 | err2=move_addr_to_user(address, msg.msg_namelen, addr, addr_len); |
| 1656 | if(err2<0) |
| 1657 | err=err2; |
| 1658 | } |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1659 | out: |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1660 | fput_light(sock_file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1661 | return err; |
| 1662 | } |
| 1663 | |
| 1664 | /* |
| 1665 | * Receive a datagram from a socket. |
| 1666 | */ |
| 1667 | |
| 1668 | asmlinkage long sys_recv(int fd, void __user * ubuf, size_t size, unsigned flags) |
| 1669 | { |
| 1670 | return sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); |
| 1671 | } |
| 1672 | |
| 1673 | /* |
| 1674 | * Set a socket option. Because we don't know the option lengths we have |
| 1675 | * to pass the user mode parameter for the protocols to sort out. |
| 1676 | */ |
| 1677 | |
| 1678 | asmlinkage long sys_setsockopt(int fd, int level, int optname, char __user *optval, int optlen) |
| 1679 | { |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1680 | int err, fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1681 | struct socket *sock; |
| 1682 | |
| 1683 | if (optlen < 0) |
| 1684 | return -EINVAL; |
| 1685 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1686 | if ((sock = sockfd_lookup_light(fd, &err, &fput_needed)) != NULL) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1687 | { |
| 1688 | err = security_socket_setsockopt(sock,level,optname); |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1689 | if (err) |
| 1690 | goto out_put; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1691 | |
| 1692 | if (level == SOL_SOCKET) |
| 1693 | err=sock_setsockopt(sock,level,optname,optval,optlen); |
| 1694 | else |
| 1695 | err=sock->ops->setsockopt(sock, level, optname, optval, optlen); |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1696 | out_put: |
| 1697 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1698 | } |
| 1699 | return err; |
| 1700 | } |
| 1701 | |
| 1702 | /* |
| 1703 | * Get a socket option. Because we don't know the option lengths we have |
| 1704 | * to pass a user mode parameter for the protocols to sort out. |
| 1705 | */ |
| 1706 | |
| 1707 | asmlinkage long sys_getsockopt(int fd, int level, int optname, char __user *optval, int __user *optlen) |
| 1708 | { |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1709 | int err, fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1710 | struct socket *sock; |
| 1711 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1712 | if ((sock = sockfd_lookup_light(fd, &err, &fput_needed)) != NULL) { |
| 1713 | err = security_socket_getsockopt(sock, level, optname); |
| 1714 | if (err) |
| 1715 | goto out_put; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1716 | |
| 1717 | if (level == SOL_SOCKET) |
| 1718 | err=sock_getsockopt(sock,level,optname,optval,optlen); |
| 1719 | else |
| 1720 | err=sock->ops->getsockopt(sock, level, optname, optval, optlen); |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1721 | out_put: |
| 1722 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1723 | } |
| 1724 | return err; |
| 1725 | } |
| 1726 | |
| 1727 | |
| 1728 | /* |
| 1729 | * Shutdown a socket. |
| 1730 | */ |
| 1731 | |
| 1732 | asmlinkage long sys_shutdown(int fd, int how) |
| 1733 | { |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1734 | int err, fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1735 | struct socket *sock; |
| 1736 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1737 | if ((sock = sockfd_lookup_light(fd, &err, &fput_needed))!=NULL) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1738 | { |
| 1739 | err = security_socket_shutdown(sock, how); |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1740 | if (!err) |
| 1741 | err = sock->ops->shutdown(sock, how); |
| 1742 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1743 | } |
| 1744 | return err; |
| 1745 | } |
| 1746 | |
| 1747 | /* A couple of helpful macros for getting the address of the 32/64 bit |
| 1748 | * fields which are the same type (int / unsigned) on our platforms. |
| 1749 | */ |
| 1750 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) |
| 1751 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) |
| 1752 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) |
| 1753 | |
| 1754 | |
| 1755 | /* |
| 1756 | * BSD sendmsg interface |
| 1757 | */ |
| 1758 | |
| 1759 | asmlinkage long sys_sendmsg(int fd, struct msghdr __user *msg, unsigned flags) |
| 1760 | { |
| 1761 | struct compat_msghdr __user *msg_compat = (struct compat_msghdr __user *)msg; |
| 1762 | struct socket *sock; |
| 1763 | char address[MAX_SOCK_ADDR]; |
| 1764 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
Alex Williamson | b9d717a | 2005-09-26 14:28:02 -0700 | [diff] [blame] | 1765 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
| 1766 | __attribute__ ((aligned (sizeof(__kernel_size_t)))); |
| 1767 | /* 20 is size of ipv6_pktinfo */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1768 | unsigned char *ctl_buf = ctl; |
| 1769 | struct msghdr msg_sys; |
| 1770 | int err, ctl_len, iov_size, total_len; |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1771 | int fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1772 | |
| 1773 | err = -EFAULT; |
| 1774 | if (MSG_CMSG_COMPAT & flags) { |
| 1775 | if (get_compat_msghdr(&msg_sys, msg_compat)) |
| 1776 | return -EFAULT; |
| 1777 | } else if (copy_from_user(&msg_sys, msg, sizeof(struct msghdr))) |
| 1778 | return -EFAULT; |
| 1779 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1780 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1781 | if (!sock) |
| 1782 | goto out; |
| 1783 | |
| 1784 | /* do not move before msg_sys is valid */ |
| 1785 | err = -EMSGSIZE; |
| 1786 | if (msg_sys.msg_iovlen > UIO_MAXIOV) |
| 1787 | goto out_put; |
| 1788 | |
| 1789 | /* Check whether to allocate the iovec area*/ |
| 1790 | err = -ENOMEM; |
| 1791 | iov_size = msg_sys.msg_iovlen * sizeof(struct iovec); |
| 1792 | if (msg_sys.msg_iovlen > UIO_FASTIOV) { |
| 1793 | iov = sock_kmalloc(sock->sk, iov_size, GFP_KERNEL); |
| 1794 | if (!iov) |
| 1795 | goto out_put; |
| 1796 | } |
| 1797 | |
| 1798 | /* This will also move the address data into kernel space */ |
| 1799 | if (MSG_CMSG_COMPAT & flags) { |
| 1800 | err = verify_compat_iovec(&msg_sys, iov, address, VERIFY_READ); |
| 1801 | } else |
| 1802 | err = verify_iovec(&msg_sys, iov, address, VERIFY_READ); |
| 1803 | if (err < 0) |
| 1804 | goto out_freeiov; |
| 1805 | total_len = err; |
| 1806 | |
| 1807 | err = -ENOBUFS; |
| 1808 | |
| 1809 | if (msg_sys.msg_controllen > INT_MAX) |
| 1810 | goto out_freeiov; |
| 1811 | ctl_len = msg_sys.msg_controllen; |
| 1812 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
Al Viro | 8920e8f | 2005-09-07 18:28:51 -0700 | [diff] [blame] | 1813 | err = cmsghdr_from_user_compat_to_kern(&msg_sys, sock->sk, ctl, sizeof(ctl)); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1814 | if (err) |
| 1815 | goto out_freeiov; |
| 1816 | ctl_buf = msg_sys.msg_control; |
Al Viro | 8920e8f | 2005-09-07 18:28:51 -0700 | [diff] [blame] | 1817 | ctl_len = msg_sys.msg_controllen; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1818 | } else if (ctl_len) { |
| 1819 | if (ctl_len > sizeof(ctl)) |
| 1820 | { |
| 1821 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
| 1822 | if (ctl_buf == NULL) |
| 1823 | goto out_freeiov; |
| 1824 | } |
| 1825 | err = -EFAULT; |
| 1826 | /* |
| 1827 | * Careful! Before this, msg_sys.msg_control contains a user pointer. |
| 1828 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
| 1829 | * checking falls down on this. |
| 1830 | */ |
| 1831 | if (copy_from_user(ctl_buf, (void __user *) msg_sys.msg_control, ctl_len)) |
| 1832 | goto out_freectl; |
| 1833 | msg_sys.msg_control = ctl_buf; |
| 1834 | } |
| 1835 | msg_sys.msg_flags = flags; |
| 1836 | |
| 1837 | if (sock->file->f_flags & O_NONBLOCK) |
| 1838 | msg_sys.msg_flags |= MSG_DONTWAIT; |
| 1839 | err = sock_sendmsg(sock, &msg_sys, total_len); |
| 1840 | |
| 1841 | out_freectl: |
| 1842 | if (ctl_buf != ctl) |
| 1843 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
| 1844 | out_freeiov: |
| 1845 | if (iov != iovstack) |
| 1846 | sock_kfree_s(sock->sk, iov, iov_size); |
| 1847 | out_put: |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1848 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1849 | out: |
| 1850 | return err; |
| 1851 | } |
| 1852 | |
| 1853 | /* |
| 1854 | * BSD recvmsg interface |
| 1855 | */ |
| 1856 | |
| 1857 | asmlinkage long sys_recvmsg(int fd, struct msghdr __user *msg, unsigned int flags) |
| 1858 | { |
| 1859 | struct compat_msghdr __user *msg_compat = (struct compat_msghdr __user *)msg; |
| 1860 | struct socket *sock; |
| 1861 | struct iovec iovstack[UIO_FASTIOV]; |
| 1862 | struct iovec *iov=iovstack; |
| 1863 | struct msghdr msg_sys; |
| 1864 | unsigned long cmsg_ptr; |
| 1865 | int err, iov_size, total_len, len; |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1866 | int fput_needed; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1867 | |
| 1868 | /* kernel mode address */ |
| 1869 | char addr[MAX_SOCK_ADDR]; |
| 1870 | |
| 1871 | /* user mode address pointers */ |
| 1872 | struct sockaddr __user *uaddr; |
| 1873 | int __user *uaddr_len; |
| 1874 | |
| 1875 | if (MSG_CMSG_COMPAT & flags) { |
| 1876 | if (get_compat_msghdr(&msg_sys, msg_compat)) |
| 1877 | return -EFAULT; |
| 1878 | } else |
| 1879 | if (copy_from_user(&msg_sys,msg,sizeof(struct msghdr))) |
| 1880 | return -EFAULT; |
| 1881 | |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1882 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1883 | if (!sock) |
| 1884 | goto out; |
| 1885 | |
| 1886 | err = -EMSGSIZE; |
| 1887 | if (msg_sys.msg_iovlen > UIO_MAXIOV) |
| 1888 | goto out_put; |
| 1889 | |
| 1890 | /* Check whether to allocate the iovec area*/ |
| 1891 | err = -ENOMEM; |
| 1892 | iov_size = msg_sys.msg_iovlen * sizeof(struct iovec); |
| 1893 | if (msg_sys.msg_iovlen > UIO_FASTIOV) { |
| 1894 | iov = sock_kmalloc(sock->sk, iov_size, GFP_KERNEL); |
| 1895 | if (!iov) |
| 1896 | goto out_put; |
| 1897 | } |
| 1898 | |
| 1899 | /* |
| 1900 | * Save the user-mode address (verify_iovec will change the |
| 1901 | * kernel msghdr to use the kernel address space) |
| 1902 | */ |
| 1903 | |
| 1904 | uaddr = (void __user *) msg_sys.msg_name; |
| 1905 | uaddr_len = COMPAT_NAMELEN(msg); |
| 1906 | if (MSG_CMSG_COMPAT & flags) { |
| 1907 | err = verify_compat_iovec(&msg_sys, iov, addr, VERIFY_WRITE); |
| 1908 | } else |
| 1909 | err = verify_iovec(&msg_sys, iov, addr, VERIFY_WRITE); |
| 1910 | if (err < 0) |
| 1911 | goto out_freeiov; |
| 1912 | total_len=err; |
| 1913 | |
| 1914 | cmsg_ptr = (unsigned long)msg_sys.msg_control; |
| 1915 | msg_sys.msg_flags = 0; |
| 1916 | if (MSG_CMSG_COMPAT & flags) |
| 1917 | msg_sys.msg_flags = MSG_CMSG_COMPAT; |
| 1918 | |
| 1919 | if (sock->file->f_flags & O_NONBLOCK) |
| 1920 | flags |= MSG_DONTWAIT; |
| 1921 | err = sock_recvmsg(sock, &msg_sys, total_len, flags); |
| 1922 | if (err < 0) |
| 1923 | goto out_freeiov; |
| 1924 | len = err; |
| 1925 | |
| 1926 | if (uaddr != NULL) { |
| 1927 | err = move_addr_to_user(addr, msg_sys.msg_namelen, uaddr, uaddr_len); |
| 1928 | if (err < 0) |
| 1929 | goto out_freeiov; |
| 1930 | } |
David S. Miller | 37f7f42 | 2005-09-16 16:51:01 -0700 | [diff] [blame] | 1931 | err = __put_user((msg_sys.msg_flags & ~MSG_CMSG_COMPAT), |
| 1932 | COMPAT_FLAGS(msg)); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1933 | if (err) |
| 1934 | goto out_freeiov; |
| 1935 | if (MSG_CMSG_COMPAT & flags) |
| 1936 | err = __put_user((unsigned long)msg_sys.msg_control-cmsg_ptr, |
| 1937 | &msg_compat->msg_controllen); |
| 1938 | else |
| 1939 | err = __put_user((unsigned long)msg_sys.msg_control-cmsg_ptr, |
| 1940 | &msg->msg_controllen); |
| 1941 | if (err) |
| 1942 | goto out_freeiov; |
| 1943 | err = len; |
| 1944 | |
| 1945 | out_freeiov: |
| 1946 | if (iov != iovstack) |
| 1947 | sock_kfree_s(sock->sk, iov, iov_size); |
| 1948 | out_put: |
Benjamin LaHaise | 6cb153c | 2006-03-20 22:27:12 -0800 | [diff] [blame] | 1949 | fput_light(sock->file, fput_needed); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1950 | out: |
| 1951 | return err; |
| 1952 | } |
| 1953 | |
| 1954 | #ifdef __ARCH_WANT_SYS_SOCKETCALL |
| 1955 | |
| 1956 | /* Argument list sizes for sys_socketcall */ |
| 1957 | #define AL(x) ((x) * sizeof(unsigned long)) |
| 1958 | static unsigned char nargs[18]={AL(0),AL(3),AL(3),AL(3),AL(2),AL(3), |
| 1959 | AL(3),AL(3),AL(4),AL(4),AL(4),AL(6), |
| 1960 | AL(6),AL(2),AL(5),AL(5),AL(3),AL(3)}; |
| 1961 | #undef AL |
| 1962 | |
| 1963 | /* |
| 1964 | * System call vectors. |
| 1965 | * |
| 1966 | * Argument checking cleaned up. Saved 20% in size. |
| 1967 | * This function doesn't need to set the kernel lock because |
| 1968 | * it is set by the callees. |
| 1969 | */ |
| 1970 | |
| 1971 | asmlinkage long sys_socketcall(int call, unsigned long __user *args) |
| 1972 | { |
| 1973 | unsigned long a[6]; |
| 1974 | unsigned long a0,a1; |
| 1975 | int err; |
| 1976 | |
| 1977 | if(call<1||call>SYS_RECVMSG) |
| 1978 | return -EINVAL; |
| 1979 | |
| 1980 | /* copy_from_user should be SMP safe. */ |
| 1981 | if (copy_from_user(a, args, nargs[call])) |
| 1982 | return -EFAULT; |
David Woodhouse | 3ec3b2f | 2005-05-17 12:08:48 +0100 | [diff] [blame] | 1983 | |
David Woodhouse | 4bcff1b | 2005-06-02 12:13:21 +0100 | [diff] [blame] | 1984 | err = audit_socketcall(nargs[call]/sizeof(unsigned long), a); |
David Woodhouse | 3ec3b2f | 2005-05-17 12:08:48 +0100 | [diff] [blame] | 1985 | if (err) |
| 1986 | return err; |
| 1987 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1988 | a0=a[0]; |
| 1989 | a1=a[1]; |
| 1990 | |
| 1991 | switch(call) |
| 1992 | { |
| 1993 | case SYS_SOCKET: |
| 1994 | err = sys_socket(a0,a1,a[2]); |
| 1995 | break; |
| 1996 | case SYS_BIND: |
| 1997 | err = sys_bind(a0,(struct sockaddr __user *)a1, a[2]); |
| 1998 | break; |
| 1999 | case SYS_CONNECT: |
| 2000 | err = sys_connect(a0, (struct sockaddr __user *)a1, a[2]); |
| 2001 | break; |
| 2002 | case SYS_LISTEN: |
| 2003 | err = sys_listen(a0,a1); |
| 2004 | break; |
| 2005 | case SYS_ACCEPT: |
| 2006 | err = sys_accept(a0,(struct sockaddr __user *)a1, (int __user *)a[2]); |
| 2007 | break; |
| 2008 | case SYS_GETSOCKNAME: |
| 2009 | err = sys_getsockname(a0,(struct sockaddr __user *)a1, (int __user *)a[2]); |
| 2010 | break; |
| 2011 | case SYS_GETPEERNAME: |
| 2012 | err = sys_getpeername(a0, (struct sockaddr __user *)a1, (int __user *)a[2]); |
| 2013 | break; |
| 2014 | case SYS_SOCKETPAIR: |
| 2015 | err = sys_socketpair(a0,a1, a[2], (int __user *)a[3]); |
| 2016 | break; |
| 2017 | case SYS_SEND: |
| 2018 | err = sys_send(a0, (void __user *)a1, a[2], a[3]); |
| 2019 | break; |
| 2020 | case SYS_SENDTO: |
| 2021 | err = sys_sendto(a0,(void __user *)a1, a[2], a[3], |
| 2022 | (struct sockaddr __user *)a[4], a[5]); |
| 2023 | break; |
| 2024 | case SYS_RECV: |
| 2025 | err = sys_recv(a0, (void __user *)a1, a[2], a[3]); |
| 2026 | break; |
| 2027 | case SYS_RECVFROM: |
| 2028 | err = sys_recvfrom(a0, (void __user *)a1, a[2], a[3], |
| 2029 | (struct sockaddr __user *)a[4], (int __user *)a[5]); |
| 2030 | break; |
| 2031 | case SYS_SHUTDOWN: |
| 2032 | err = sys_shutdown(a0,a1); |
| 2033 | break; |
| 2034 | case SYS_SETSOCKOPT: |
| 2035 | err = sys_setsockopt(a0, a1, a[2], (char __user *)a[3], a[4]); |
| 2036 | break; |
| 2037 | case SYS_GETSOCKOPT: |
| 2038 | err = sys_getsockopt(a0, a1, a[2], (char __user *)a[3], (int __user *)a[4]); |
| 2039 | break; |
| 2040 | case SYS_SENDMSG: |
| 2041 | err = sys_sendmsg(a0, (struct msghdr __user *) a1, a[2]); |
| 2042 | break; |
| 2043 | case SYS_RECVMSG: |
| 2044 | err = sys_recvmsg(a0, (struct msghdr __user *) a1, a[2]); |
| 2045 | break; |
| 2046 | default: |
| 2047 | err = -EINVAL; |
| 2048 | break; |
| 2049 | } |
| 2050 | return err; |
| 2051 | } |
| 2052 | |
| 2053 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
| 2054 | |
| 2055 | /* |
| 2056 | * This function is called by a protocol handler that wants to |
| 2057 | * advertise its address family, and have it linked into the |
| 2058 | * SOCKET module. |
| 2059 | */ |
| 2060 | |
| 2061 | int sock_register(struct net_proto_family *ops) |
| 2062 | { |
| 2063 | int err; |
| 2064 | |
| 2065 | if (ops->family >= NPROTO) { |
| 2066 | printk(KERN_CRIT "protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
| 2067 | return -ENOBUFS; |
| 2068 | } |
| 2069 | net_family_write_lock(); |
| 2070 | err = -EEXIST; |
| 2071 | if (net_families[ops->family] == NULL) { |
| 2072 | net_families[ops->family]=ops; |
| 2073 | err = 0; |
| 2074 | } |
| 2075 | net_family_write_unlock(); |
| 2076 | printk(KERN_INFO "NET: Registered protocol family %d\n", |
| 2077 | ops->family); |
| 2078 | return err; |
| 2079 | } |
| 2080 | |
| 2081 | /* |
| 2082 | * This function is called by a protocol handler that wants to |
| 2083 | * remove its address family, and have it unlinked from the |
| 2084 | * SOCKET module. |
| 2085 | */ |
| 2086 | |
| 2087 | int sock_unregister(int family) |
| 2088 | { |
| 2089 | if (family < 0 || family >= NPROTO) |
| 2090 | return -1; |
| 2091 | |
| 2092 | net_family_write_lock(); |
| 2093 | net_families[family]=NULL; |
| 2094 | net_family_write_unlock(); |
| 2095 | printk(KERN_INFO "NET: Unregistered protocol family %d\n", |
| 2096 | family); |
| 2097 | return 0; |
| 2098 | } |
| 2099 | |
Andi Kleen | 77d76ea | 2005-12-22 12:43:42 -0800 | [diff] [blame] | 2100 | static int __init sock_init(void) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2101 | { |
| 2102 | /* |
| 2103 | * Initialize sock SLAB cache. |
| 2104 | */ |
| 2105 | |
| 2106 | sk_init(); |
| 2107 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2108 | /* |
| 2109 | * Initialize skbuff SLAB cache |
| 2110 | */ |
| 2111 | skb_init(); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2112 | |
| 2113 | /* |
| 2114 | * Initialize the protocols module. |
| 2115 | */ |
| 2116 | |
| 2117 | init_inodecache(); |
| 2118 | register_filesystem(&sock_fs_type); |
| 2119 | sock_mnt = kern_mount(&sock_fs_type); |
Andi Kleen | 77d76ea | 2005-12-22 12:43:42 -0800 | [diff] [blame] | 2120 | |
| 2121 | /* The real protocol initialization is performed in later initcalls. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2122 | */ |
| 2123 | |
| 2124 | #ifdef CONFIG_NETFILTER |
| 2125 | netfilter_init(); |
| 2126 | #endif |
David S. Miller | cbeb321 | 2005-12-22 12:58:55 -0800 | [diff] [blame] | 2127 | |
| 2128 | return 0; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2129 | } |
| 2130 | |
Andi Kleen | 77d76ea | 2005-12-22 12:43:42 -0800 | [diff] [blame] | 2131 | core_initcall(sock_init); /* early initcall */ |
| 2132 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2133 | #ifdef CONFIG_PROC_FS |
| 2134 | void socket_seq_show(struct seq_file *seq) |
| 2135 | { |
| 2136 | int cpu; |
| 2137 | int counter = 0; |
| 2138 | |
KAMEZAWA Hiroyuki | 6f91204 | 2006-04-10 22:52:50 -0700 | [diff] [blame] | 2139 | for_each_possible_cpu(cpu) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2140 | counter += per_cpu(sockets_in_use, cpu); |
| 2141 | |
| 2142 | /* It can be negative, by the way. 8) */ |
| 2143 | if (counter < 0) |
| 2144 | counter = 0; |
| 2145 | |
| 2146 | seq_printf(seq, "sockets: used %d\n", counter); |
| 2147 | } |
| 2148 | #endif /* CONFIG_PROC_FS */ |
| 2149 | |
Shaun Pereira | 89bbfc9 | 2006-03-21 23:58:08 -0800 | [diff] [blame] | 2150 | #ifdef CONFIG_COMPAT |
| 2151 | static long compat_sock_ioctl(struct file *file, unsigned cmd, |
| 2152 | unsigned long arg) |
| 2153 | { |
| 2154 | struct socket *sock = file->private_data; |
| 2155 | int ret = -ENOIOCTLCMD; |
| 2156 | |
| 2157 | if (sock->ops->compat_ioctl) |
| 2158 | ret = sock->ops->compat_ioctl(sock, cmd, arg); |
| 2159 | |
| 2160 | return ret; |
| 2161 | } |
| 2162 | #endif |
| 2163 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2164 | /* ABI emulation layers need these two */ |
| 2165 | EXPORT_SYMBOL(move_addr_to_kernel); |
| 2166 | EXPORT_SYMBOL(move_addr_to_user); |
| 2167 | EXPORT_SYMBOL(sock_create); |
| 2168 | EXPORT_SYMBOL(sock_create_kern); |
| 2169 | EXPORT_SYMBOL(sock_create_lite); |
| 2170 | EXPORT_SYMBOL(sock_map_fd); |
| 2171 | EXPORT_SYMBOL(sock_recvmsg); |
| 2172 | EXPORT_SYMBOL(sock_register); |
| 2173 | EXPORT_SYMBOL(sock_release); |
| 2174 | EXPORT_SYMBOL(sock_sendmsg); |
| 2175 | EXPORT_SYMBOL(sock_unregister); |
| 2176 | EXPORT_SYMBOL(sock_wake_async); |
| 2177 | EXPORT_SYMBOL(sockfd_lookup); |
| 2178 | EXPORT_SYMBOL(kernel_sendmsg); |
| 2179 | EXPORT_SYMBOL(kernel_recvmsg); |