/*
* INET An implementation of the TCP/IP protocol suite for the LINUX
* operating system. INET is implemented using the BSD Socket
* interface as the means of communication with the user level.
*
* Generic socket support routines. Memory allocators, socket lock/release
* handler for protocols to use and generic option handler.
*
*
* Version: $Id: sock.c,v 1.102 2000/12/11 23:00:24 davem Exp $
*
* Authors: Ross Biro, <bir7@leland.Stanford.Edu>
* Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
* Florian La Roche, <flla@stud.uni-sb.de>
* Alan Cox, <A.Cox@swansea.ac.uk>
*
* Fixes:
* Alan Cox : Numerous verify_area() problems
* Alan Cox : Connecting on a connecting socket
* now returns an error for tcp.
* Alan Cox : sock->protocol is set correctly.
* and is not sometimes left as 0.
* Alan Cox : connect handles icmp errors on a
* connect properly. Unfortunately there
* is a restart syscall nasty there. I
* can't match BSD without hacking the C
* library. Ideas urgently sought!
* Alan Cox : Disallow bind() to addresses that are
* not ours - especially broadcast ones!!
* Alan Cox : Socket 1024 _IS_ ok for users. (fencepost)
* Alan Cox : sock_wfree/sock_rfree don't destroy sockets,
* instead they leave that for the DESTROY timer.
* Alan Cox : Clean up error flag in accept
* Alan Cox : TCP ack handling is buggy, the DESTROY timer
* was buggy. Put a remove_sock() in the handler
* for memory when we hit 0. Also altered the timer
* code. The ACK stuff can wait and needs major
* TCP layer surgery.
* Alan Cox : Fixed TCP ack bug, removed remove sock
* and fixed timer/inet_bh race.
* Alan Cox : Added zapped flag for TCP
* Alan Cox : Move kfree_skb into skbuff.c and tidied up surplus code
* Alan Cox : for new sk_buff allocations wmalloc/rmalloc now call alloc_skb
* Alan Cox : kfree_s calls now are kfree_skbmem so we can track skb resources
* Alan Cox : Supports socket option broadcast now as does udp. Packet and raw need fixing.
* Alan Cox : Added RCVBUF,SNDBUF size setting. It suddenly occurred to me how easy it was so...
* Rick Sladkey : Relaxed UDP rules for matching packets.
* C.E.Hawkins : IFF_PROMISC/SIOCGHWADDR support
* Pauline Middelink : identd support
* Alan Cox : Fixed connect() taking signals I think.
* Alan Cox : SO_LINGER supported
* Alan Cox : Error reporting fixes
* Anonymous : inet_create tidied up (sk->reuse setting)
* Alan Cox : inet sockets don't set sk->type!
* Alan Cox : Split socket option code
* Alan Cox : Callbacks
* Alan Cox : Nagle flag for Charles & Johannes stuff
* Alex : Removed restriction on inet fioctl
* Alan Cox : Splitting INET from NET core
* Alan Cox : Fixed bogus SO_TYPE handling in getsockopt()
* Adam Caldwell : Missing return in SO_DONTROUTE/SO_DEBUG code
* Alan Cox : Split IP from generic code
* Alan Cox : New kfree_skbmem()
* Alan Cox : Make SO_DEBUG superuser only.
* Alan Cox : Allow anyone to clear SO_DEBUG
* (compatibility fix)
* Alan Cox : Added optimistic memory grabbing for AF_UNIX throughput.
* Alan Cox : Allocator for a socket is settable.
* Alan Cox : SO_ERROR includes soft errors.
* Alan Cox : Allow NULL arguments on some SO_ opts
* Alan Cox : Generic socket allocation to make hooks
* easier (suggested by Craig Metz).
* Michael Pall : SO_ERROR returns positive errno again
* Steve Whitehouse: Added default destructor to free
* protocol private data.
* Steve Whitehouse: Added various other default routines
* common to several socket families.
* Chris Evans : Call suser() check last on F_SETOWN
* Jay Schulist : Added SO_ATTACH_FILTER and SO_DETACH_FILTER.
* Andi Kleen : Add sock_kmalloc()/sock_kfree_s()
* Andi Kleen : Fix write_space callback
*
* To Fix:
*
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version
* 2 of the License, or (at your option) any later version.
*/
#include <linux/config.h>
#include <linux/errno.h>
#include <linux/types.h>
#include <linux/socket.h>
#include <linux/in.h>
#include <linux/kernel.h>
#include <linux/major.h>
#include <linux/sched.h>
#include <linux/timer.h>
#include <linux/string.h>
#include <linux/sockios.h>
#include <linux/net.h>
#include <linux/fcntl.h>
#include <linux/mm.h>
#include <linux/slab.h>
#include <linux/interrupt.h>
#include <linux/poll.h>
#include <linux/init.h>
#include <asm/uaccess.h>
#include <asm/system.h>
#include <linux/inet.h>
#include <linux/netdevice.h>
#include <net/ip.h>
#include <net/protocol.h>
#include <net/arp.h>
#include <net/route.h>
#include <net/tcp.h>
#include <net/udp.h>
#include <linux/skbuff.h>
#include <net/sock.h>
#include <net/raw.h>
#include <net/icmp.h>
#include <linux/ipsec.h>
#ifdef CONFIG_FILTER
#include <linux/filter.h>
#endif
#define min(a,b) ((a)<(b)?(a):(b))
/* Run time adjustable parameters. */
__u32 sysctl_wmem_max = SK_WMEM_MAX;
__u32 sysctl_rmem_max = SK_RMEM_MAX;
__u32 sysctl_wmem_default = SK_WMEM_MAX;
__u32 sysctl_rmem_default = SK_RMEM_MAX;
/* Maximal space eaten by iovec or ancilliary data plus some space */
int sysctl_optmem_max = sizeof(unsigned long)*(2*UIO_MAXIOV + 512);
143 static int sock_set_timeout(long *timeo_p, char *optval, int optlen)
{
struct timeval tv;
147 if (optlen < sizeof(tv))
148 return -EINVAL;
149 if (copy_from_user(&tv, optval, sizeof(tv)))
150 return -EFAULT;
*timeo_p = MAX_SCHEDULE_TIMEOUT;
153 if (tv.tv_sec == 0 && tv.tv_usec == 0)
154 return 0;
155 if (tv.tv_sec < (MAX_SCHEDULE_TIMEOUT/HZ - 1))
*timeo_p = tv.tv_sec*HZ + (tv.tv_usec+(1000000/HZ-1))/(1000000/HZ);
157 return 0;
}
/*
* This is meant for all protocols to use and covers goings on
* at the socket level. Everything here is generic.
*/
165 int sock_setsockopt(struct socket *sock, int level, int optname,
char *optval, int optlen)
{
struct sock *sk=sock->sk;
#ifdef CONFIG_FILTER
struct sk_filter *filter;
#endif
int val;
int valbool;
int err;
struct linger ling;
int ret = 0;
/*
* Options without arguments
*/
#ifdef SO_DONTLINGER /* Compatibility item... */
switch(optname)
{
case SO_DONTLINGER:
sk->linger=0;
return 0;
}
#endif
191 if(optlen<sizeof(int))
192 return(-EINVAL);
err = get_user(val, (int *)optval);
195 if (err)
196 return err;
valbool = val?1:0;
200 lock_sock(sk);
202 switch(optname)
{
204 case SO_DEBUG:
205 if(val && !capable(CAP_NET_ADMIN))
{
ret = -EACCES;
}
209 else
sk->debug=valbool;
211 break;
212 case SO_REUSEADDR:
sk->reuse = valbool;
214 break;
215 case SO_TYPE:
216 case SO_ERROR:
ret = -ENOPROTOOPT;
218 break;
219 case SO_DONTROUTE:
sk->localroute=valbool;
221 break;
222 case SO_BROADCAST:
sk->broadcast=valbool;
224 break;
225 case SO_SNDBUF:
/* Don't error on this BSD doesn't and if you think
about it this is right. Otherwise apps have to
play 'guess the biggest size' games. RCVBUF/SNDBUF
are treated in BSD as hints */
231 if (val > sysctl_wmem_max)
val = sysctl_wmem_max;
sk->userlocks |= SOCK_SNDBUF_LOCK;
sk->sndbuf = max(val*2,SOCK_MIN_SNDBUF);
/*
* Wake up sending tasks if we
* upped the value.
*/
sk->write_space(sk);
242 break;
244 case SO_RCVBUF:
/* Don't error on this BSD doesn't and if you think
about it this is right. Otherwise apps have to
play 'guess the biggest size' games. RCVBUF/SNDBUF
are treated in BSD as hints */
250 if (val > sysctl_rmem_max)
val = sysctl_rmem_max;
sk->userlocks |= SOCK_RCVBUF_LOCK;
/* FIXME: is this lower bound the right one? */
sk->rcvbuf = max(val*2,SOCK_MIN_RCVBUF);
256 break;
258 case SO_KEEPALIVE:
#ifdef CONFIG_INET
260 if (sk->protocol == IPPROTO_TCP)
{
tcp_set_keepalive(sk, valbool);
}
#endif
sk->keepopen = valbool;
266 break;
268 case SO_OOBINLINE:
sk->urginline = valbool;
270 break;
272 case SO_NO_CHECK:
sk->no_check = valbool;
274 break;
276 case SO_PRIORITY:
277 if ((val >= 0 && val <= 6) || capable(CAP_NET_ADMIN))
sk->priority = val;
279 else
ret = -EPERM;
281 break;
283 case SO_LINGER:
284 if(optlen<sizeof(ling)) {
ret = -EINVAL; /* 1003.1g */
286 break;
}
288 if (copy_from_user(&ling,optval,sizeof(ling))) {
ret = -EFAULT;
290 break;
}
292 if(ling.l_onoff==0) {
sk->linger=0;
294 } else {
#if (BITS_PER_LONG == 32)
296 if (ling.l_linger >= MAX_SCHEDULE_TIMEOUT/HZ)
sk->lingertime=MAX_SCHEDULE_TIMEOUT;
298 else
#endif
sk->lingertime=ling.l_linger*HZ;
sk->linger=1;
}
303 break;
305 case SO_BSDCOMPAT:
sk->bsdism = valbool;
307 break;
309 case SO_PASSCRED:
sock->passcred = valbool;
311 break;
313 case SO_TIMESTAMP:
sk->rcvtstamp = valbool;
315 break;
317 case SO_RCVLOWAT:
318 if (val < 0)
val = INT_MAX;
sk->rcvlowat = val ? : 1;
321 break;
323 case SO_RCVTIMEO:
ret = sock_set_timeout(&sk->rcvtimeo, optval, optlen);
325 break;
327 case SO_SNDTIMEO:
ret = sock_set_timeout(&sk->sndtimeo, optval, optlen);
329 break;
#ifdef CONFIG_NETDEVICES
332 case SO_BINDTODEVICE:
{
char devname[IFNAMSIZ];
/* Sorry... */
337 if (!capable(CAP_NET_RAW)) {
ret = -EPERM;
339 break;
}
/* Bind this socket to a particular device like "eth0",
* as specified in the passed interface name. If the
* name is "" or the option length is zero the socket
* is not bound.
*/
348 if (!valbool) {
sk->bound_dev_if = 0;
350 } else {
351 if (optlen > IFNAMSIZ)
optlen = IFNAMSIZ;
353 if (copy_from_user(devname, optval, optlen)) {
ret = -EFAULT;
355 break;
}
/* Remove any cached route for this socket. */
sk_dst_reset(sk);
361 if (devname[0] == '\0') {
sk->bound_dev_if = 0;
363 } else {
struct net_device *dev = dev_get_by_name(devname);
365 if (!dev) {
ret = -ENODEV;
367 break;
}
sk->bound_dev_if = dev->ifindex;
dev_put(dev);
}
}
373 break;
}
#endif
#ifdef CONFIG_FILTER
case SO_ATTACH_FILTER:
ret = -EINVAL;
if (optlen == sizeof(struct sock_fprog)) {
struct sock_fprog fprog;
ret = -EFAULT;
if (copy_from_user(&fprog, optval, sizeof(fprog)))
break;
ret = sk_attach_filter(&fprog, sk);
}
break;
case SO_DETACH_FILTER:
spin_lock_bh(&sk->lock.slock);
filter = sk->filter;
if (filter) {
sk->filter = NULL;
spin_unlock_bh(&sk->lock.slock);
sk_filter_release(sk, filter);
break;
}
spin_unlock_bh(&sk->lock.slock);
ret = -ENONET;
break;
#endif
/* We implement the SO_SNDLOWAT etc to
not be settable (1003.1g 5.3) */
407 default:
ret = -ENOPROTOOPT;
409 break;
}
411 release_sock(sk);
412 return ret;
}
416 int sock_getsockopt(struct socket *sock, int level, int optname,
char *optval, int *optlen)
{
struct sock *sk = sock->sk;
union
{
int val;
struct linger ling;
struct timeval tm;
} v;
int lv=sizeof(int),len;
430 if(get_user(len,optlen))
431 return -EFAULT;
433 switch(optname)
{
435 case SO_DEBUG:
v.val = sk->debug;
437 break;
439 case SO_DONTROUTE:
v.val = sk->localroute;
441 break;
443 case SO_BROADCAST:
v.val= sk->broadcast;
445 break;
447 case SO_SNDBUF:
v.val=sk->sndbuf;
449 break;
451 case SO_RCVBUF:
v.val =sk->rcvbuf;
453 break;
455 case SO_REUSEADDR:
v.val = sk->reuse;
457 break;
459 case SO_KEEPALIVE:
v.val = sk->keepopen;
461 break;
463 case SO_TYPE:
v.val = sk->type;
465 break;
467 case SO_ERROR:
v.val = -sock_error(sk);
469 if(v.val==0)
v.val=xchg(&sk->err_soft,0);
471 break;
473 case SO_OOBINLINE:
v.val = sk->urginline;
475 break;
477 case SO_NO_CHECK:
v.val = sk->no_check;
479 break;
481 case SO_PRIORITY:
v.val = sk->priority;
483 break;
485 case SO_LINGER:
lv=sizeof(v.ling);
v.ling.l_onoff=sk->linger;
v.ling.l_linger=sk->lingertime/HZ;
489 break;
491 case SO_BSDCOMPAT:
v.val = sk->bsdism;
493 break;
495 case SO_TIMESTAMP:
v.val = sk->rcvtstamp;
497 break;
499 case SO_RCVTIMEO:
lv=sizeof(struct timeval);
501 if (sk->rcvtimeo == MAX_SCHEDULE_TIMEOUT) {
v.tm.tv_sec = 0;
v.tm.tv_usec = 0;
504 } else {
v.tm.tv_sec = sk->rcvtimeo/HZ;
v.tm.tv_usec = ((sk->rcvtimeo%HZ)*1000)/HZ;
}
508 break;
510 case SO_SNDTIMEO:
lv=sizeof(struct timeval);
512 if (sk->sndtimeo == MAX_SCHEDULE_TIMEOUT) {
v.tm.tv_sec = 0;
v.tm.tv_usec = 0;
515 } else {
v.tm.tv_sec = sk->sndtimeo/HZ;
v.tm.tv_usec = ((sk->sndtimeo%HZ)*1000)/HZ;
}
519 break;
521 case SO_RCVLOWAT:
v.val = sk->rcvlowat;
523 break;
525 case SO_SNDLOWAT:
v.val=1;
527 break;
529 case SO_PASSCRED:
v.val = sock->passcred;
531 break;
533 case SO_PEERCRED:
lv=sizeof(sk->peercred);
len=min(len, lv);
536 if(copy_to_user((void*)optval, &sk->peercred, len))
537 return -EFAULT;
538 goto lenout;
540 case SO_PEERNAME:
{
char address[128];
544 if (sock->ops->getname(sock, (struct sockaddr *)address, &lv, 2))
545 return -ENOTCONN;
546 if (lv < len)
547 return -EINVAL;
548 if(copy_to_user((void*)optval, address, len))
549 return -EFAULT;
550 goto lenout;
}
553 default:
554 return(-ENOPROTOOPT);
}
len=min(len,lv);
557 if(copy_to_user(optval,&v,len))
558 return -EFAULT;
lenout:
560 if(put_user(len, optlen))
561 return -EFAULT;
562 return 0;
}
static kmem_cache_t *sk_cachep;
/*
* All socket objects are allocated here. This is for future
* usage.
*/
572 struct sock *sk_alloc(int family, int priority, int zero_it)
{
struct sock *sk = kmem_cache_alloc(sk_cachep, priority);
576 if(sk && zero_it) {
memset(sk, 0, sizeof(struct sock));
sk->family = family;
579 sock_lock_init(sk);
}
582 return sk;
}
585 void sk_free(struct sock *sk)
{
#ifdef CONFIG_FILTER
struct sk_filter *filter;
#endif
591 if (sk->destruct)
sk->destruct(sk);
#ifdef CONFIG_FILTER
filter = sk->filter;
if (filter) {
sk_filter_release(sk, filter);
sk->filter = NULL;
}
#endif
602 if (atomic_read(&sk->omem_alloc))
printk(KERN_DEBUG "sk_free: optmem leakage (%d bytes) detected.\n", atomic_read(&sk->omem_alloc));
kmem_cache_free(sk_cachep, sk);
}
608 void __init sk_init(void)
{
sk_cachep = kmem_cache_create("sock", sizeof(struct sock), 0,
SLAB_HWCACHE_ALIGN, 0, 0);
612 if (!sk_cachep)
printk(KERN_CRIT "sk_init: Cannot create sock SLAB cache!");
615 if (num_physpages <= 4096) {
sysctl_wmem_max = 32767;
sysctl_rmem_max = 32767;
sysctl_wmem_default = 32767;
sysctl_wmem_default = 32767;
620 } else if (num_physpages >= 131072) {
sysctl_wmem_max = 131071;
sysctl_rmem_max = 131071;
}
}
/*
* Simple resource managers for sockets.
*/
/*
* Write buffer destructor automatically called from kfree_skb.
*/
634 void sock_wfree(struct sk_buff *skb)
{
struct sock *sk = skb->sk;
/* In case it might be waiting for more memory. */
atomic_sub(skb->truesize, &sk->wmem_alloc);
sk->write_space(sk);
sock_put(sk);
}
/*
* Read buffer destructor automatically called from kfree_skb.
*/
647 void sock_rfree(struct sk_buff *skb)
{
struct sock *sk = skb->sk;
atomic_sub(skb->truesize, &sk->rmem_alloc);
}
/*
* Allocate a skb from the socket's send buffer.
*/
657 struct sk_buff *sock_wmalloc(struct sock *sk, unsigned long size, int force, int priority)
{
659 if (force || atomic_read(&sk->wmem_alloc) < sk->sndbuf) {
struct sk_buff * skb = alloc_skb(size, priority);
661 if (skb) {
skb_set_owner_w(skb, sk);
663 return skb;
}
}
666 return NULL;
}
/*
* Allocate a skb from the socket's receive buffer.
*/
672 struct sk_buff *sock_rmalloc(struct sock *sk, unsigned long size, int force, int priority)
{
674 if (force || atomic_read(&sk->rmem_alloc) < sk->rcvbuf) {
struct sk_buff *skb = alloc_skb(size, priority);
676 if (skb) {
skb_set_owner_r(skb, sk);
678 return skb;
}
}
681 return NULL;
}
/*
* Allocate a memory block from the socket's option memory buffer.
*/
687 void *sock_kmalloc(struct sock *sk, int size, int priority)
{
if ((unsigned)size <= sysctl_optmem_max &&
690 atomic_read(&sk->omem_alloc)+size < sysctl_optmem_max) {
void *mem;
/* First do the add, to avoid the race if kmalloc
* might sleep.
*/
atomic_add(size, &sk->omem_alloc);
mem = kmalloc(size, priority);
697 if (mem)
698 return mem;
atomic_sub(size, &sk->omem_alloc);
}
701 return NULL;
}
/*
* Free an option memory block.
*/
707 void sock_kfree_s(struct sock *sk, void *mem, int size)
{
kfree(mem);
atomic_sub(size, &sk->omem_alloc);
}
/* It is almost wait_for_tcp_memory minus release_sock/lock_sock.
I think, these locks should be removed for datagram sockets.
*/
716 static long sock_wait_for_wmem(struct sock * sk, long timeo)
{
DECLARE_WAITQUEUE(wait, current);
clear_bit(SOCK_ASYNC_NOSPACE, &sk->socket->flags);
add_wait_queue(sk->sleep, &wait);
722 for (;;) {
723 if (signal_pending(current))
724 break;
set_bit(SOCK_NOSPACE, &sk->socket->flags);
726 set_current_state(TASK_INTERRUPTIBLE);
727 if (atomic_read(&sk->wmem_alloc) < sk->sndbuf)
728 break;
729 if (sk->shutdown & SEND_SHUTDOWN)
730 break;
731 if (sk->err)
732 break;
timeo = schedule_timeout(timeo);
}
735 __set_current_state(TASK_RUNNING);
remove_wait_queue(sk->sleep, &wait);
737 return timeo;
}
/*
* Generic send/receive buffer handlers
*/
745 struct sk_buff *sock_alloc_send_skb(struct sock *sk, unsigned long size,
unsigned long fallback, int noblock, int *errcode)
{
int err;
struct sk_buff *skb;
long timeo;
timeo = sock_sndtimeo(sk, noblock);
754 while (1) {
unsigned long try_size = size;
err = sock_error(sk);
758 if (err != 0)
759 goto failure;
/*
* We should send SIGPIPE in these cases according to
* 1003.1g draft 6.4. If we (the user) did a shutdown()
* call however we should not.
*
* Note: This routine isnt just used for datagrams and
* anyway some datagram protocols have a notion of
* close down.
*/
err = -EPIPE;
772 if (sk->shutdown&SEND_SHUTDOWN)
773 goto failure;
775 if (atomic_read(&sk->wmem_alloc) < sk->sndbuf) {
776 if (fallback) {
/* The buffer get won't block, or use the atomic queue.
* It does produce annoying no free page messages still.
*/
skb = alloc_skb(size, GFP_BUFFER);
781 if (skb)
782 break;
try_size = fallback;
}
skb = alloc_skb(try_size, sk->allocation);
786 if (skb)
787 break;
err = -ENOBUFS;
789 goto failure;
}
/*
* This means we have too many buffers for this socket already.
*/
set_bit(SOCK_ASYNC_NOSPACE, &sk->socket->flags);
set_bit(SOCK_NOSPACE, &sk->socket->flags);
err = -EAGAIN;
799 if (!timeo)
800 goto failure;
801 if (signal_pending(current))
802 goto interrupted;
timeo = sock_wait_for_wmem(sk, timeo);
}
skb_set_owner_w(skb, sk);
807 return skb;
interrupted:
err = sock_intr_errno(timeo);
failure:
*errcode = err;
813 return NULL;
}
816 void __lock_sock(struct sock *sk)
{
DECLARE_WAITQUEUE(wait, current);
add_wait_queue_exclusive(&sk->lock.wq, &wait);
821 for(;;) {
current->state = TASK_UNINTERRUPTIBLE;
823 spin_unlock_bh(&sk->lock.slock);
schedule();
825 spin_lock_bh(&sk->lock.slock);
826 if(!sk->lock.users)
827 break;
}
current->state = TASK_RUNNING;
remove_wait_queue(&sk->lock.wq, &wait);
}
833 void __release_sock(struct sock *sk)
{
struct sk_buff *skb = sk->backlog.head;
837 do {
sk->backlog.head = sk->backlog.tail = NULL;
839 bh_unlock_sock(sk);
841 do {
struct sk_buff *next = skb->next;
skb->next = NULL;
sk->backlog_rcv(sk, skb);
skb = next;
847 } while (skb != NULL);
bh_lock_sock(sk);
850 } while((skb = sk->backlog.head) != NULL);
}
/*
* Generic socket manager library. Most simpler socket families
* use this to manage their socket lists. At some point we should
* hash these. By making this generic we get the lot hashed for free.
*
* It is broken by design. All the protocols using it must be fixed. --ANK
*/
rwlock_t net_big_sklist_lock = RW_LOCK_UNLOCKED;
863 void sklist_remove_socket(struct sock **list, struct sock *sk)
{
struct sock *s;
867 write_lock_bh(&net_big_sklist_lock);
869 while ((s = *list) != NULL) {
870 if (s == sk) {
*list = s->next;
872 break;
}
list = &s->next;
}
877 write_unlock_bh(&net_big_sklist_lock);
878 if (s)
sock_put(s);
}
882 void sklist_insert_socket(struct sock **list, struct sock *sk)
{
884 write_lock_bh(&net_big_sklist_lock);
sk->next= *list;
*list=sk;
sock_hold(sk);
888 write_unlock_bh(&net_big_sklist_lock);
}
/*
* This is only called from user mode. Thus it protects itself against
* interrupt users but doesn't worry about being called during work.
* Once it is removed from the queue no interrupt or bottom half will
* touch it and we are (fairly 8-) ) safe.
*/
void sklist_destroy_socket(struct sock **list, struct sock *sk);
/*
* Handler for deferred kills.
*/
904 static void sklist_destroy_timer(unsigned long data)
{
struct sock *sk=(struct sock *)data;
sklist_destroy_socket(NULL,sk);
}
/*
* Destroy a socket. We pass NULL for a list if we know the
* socket is not on a list.
*/
915 void sklist_destroy_socket(struct sock **list,struct sock *sk)
{
struct sk_buff *skb;
918 if(list)
sklist_remove_socket(list, sk);
921 while((skb=skb_dequeue(&sk->receive_queue))!=NULL)
{
kfree_skb(skb);
}
if(atomic_read(&sk->wmem_alloc) == 0 &&
atomic_read(&sk->rmem_alloc) == 0 &&
928 sk->dead)
{
sock_put(sk);
}
932 else
{
/*
* Someone is using our buffers still.. defer
*/
init_timer(&sk->timer);
sk->timer.expires=jiffies+SOCK_DESTROY_TIME;
sk->timer.function=sklist_destroy_timer;
sk->timer.data = (unsigned long)sk;
add_timer(&sk->timer);
}
}
/*
* Set of default routines for initialising struct proto_ops when
* the protocol does not support a particular function. In certain
* cases where it makes no sense for a protocol to have a "do nothing"
* function, some default processing is provided.
*/
952 int sock_no_release(struct socket *sock)
{
954 return 0;
}
957 int sock_no_bind(struct socket *sock, struct sockaddr *saddr, int len)
{
959 return -EOPNOTSUPP;
}
962 int sock_no_connect(struct socket *sock, struct sockaddr *saddr,
int len, int flags)
{
965 return -EOPNOTSUPP;
}
968 int sock_no_socketpair(struct socket *sock1, struct socket *sock2)
{
970 return -EOPNOTSUPP;
}
973 int sock_no_accept(struct socket *sock, struct socket *newsock, int flags)
{
975 return -EOPNOTSUPP;
}
978 int sock_no_getname(struct socket *sock, struct sockaddr *saddr,
int *len, int peer)
{
981 return -EOPNOTSUPP;
}
984 unsigned int sock_no_poll(struct file * file, struct socket *sock, poll_table *pt)
{
986 return 0;
}
989 int sock_no_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
{
991 return -EOPNOTSUPP;
}
994 int sock_no_listen(struct socket *sock, int backlog)
{
996 return -EOPNOTSUPP;
}
999 int sock_no_shutdown(struct socket *sock, int how)
{
1001 return -EOPNOTSUPP;
}
1004 int sock_no_setsockopt(struct socket *sock, int level, int optname,
char *optval, int optlen)
{
1007 return -EOPNOTSUPP;
}
1010 int sock_no_getsockopt(struct socket *sock, int level, int optname,
char *optval, int *optlen)
{
1013 return -EOPNOTSUPP;
}
/*
* Note: if you add something that sleeps here then change sock_fcntl()
* to do proper fd locking.
*/
1020 int sock_no_fcntl(struct socket *sock, unsigned int cmd, unsigned long arg)
{
struct sock *sk = sock->sk;
1024 switch(cmd)
{
1026 case F_SETOWN:
/*
* This is a little restrictive, but it's the only
* way to make sure that you can't send a sigurg to
* another process.
*/
if (current->pgrp != -arg &&
current->pid != arg &&
1034 !capable(CAP_KILL)) return(-EPERM);
sk->proc = arg;
1036 return(0);
1037 case F_GETOWN:
1038 return(sk->proc);
1039 default:
1040 return(-EINVAL);
}
}
1044 int sock_no_sendmsg(struct socket *sock, struct msghdr *m, int flags,
struct scm_cookie *scm)
{
1047 return -EOPNOTSUPP;
}
1050 int sock_no_recvmsg(struct socket *sock, struct msghdr *m, int len, int flags,
struct scm_cookie *scm)
{
1053 return -EOPNOTSUPP;
}
1056 int sock_no_mmap(struct file *file, struct socket *sock, struct vm_area_struct *vma)
{
/* Mirror missing mmap method error code */
1059 return -ENODEV;
}
/*
* Default Socket Callbacks
*/
1066 void sock_def_wakeup(struct sock *sk)
{
read_lock(&sk->callback_lock);
1069 if (sk->sleep && waitqueue_active(sk->sleep))
wake_up_interruptible_all(sk->sleep);
1071 read_unlock(&sk->callback_lock);
}
1074 void sock_def_error_report(struct sock *sk)
{
read_lock(&sk->callback_lock);
1077 if (sk->sleep && waitqueue_active(sk->sleep))
wake_up_interruptible(sk->sleep);
sk_wake_async(sk,0,POLL_ERR);
1080 read_unlock(&sk->callback_lock);
}
1083 void sock_def_readable(struct sock *sk, int len)
{
read_lock(&sk->callback_lock);
1086 if (sk->sleep && waitqueue_active(sk->sleep))
wake_up_interruptible(sk->sleep);
sk_wake_async(sk,1,POLL_IN);
1089 read_unlock(&sk->callback_lock);
}
1092 void sock_def_write_space(struct sock *sk)
{
read_lock(&sk->callback_lock);
/* Do not wake up a writer until he can make "significant"
* progress. --DaveM
*/
1099 if((atomic_read(&sk->wmem_alloc) << 1) <= sk->sndbuf) {
1100 if (sk->sleep && waitqueue_active(sk->sleep))
wake_up_interruptible(sk->sleep);
/* Should agree with poll, otherwise some programs break */
1104 if (sock_writeable(sk))
sk_wake_async(sk, 2, POLL_OUT);
}
1108 read_unlock(&sk->callback_lock);
}
1111 void sock_def_destruct(struct sock *sk)
{
1113 if (sk->protinfo.destruct_hook)
kfree(sk->protinfo.destruct_hook);
}
1117 void sock_init_data(struct socket *sock, struct sock *sk)
{
skb_queue_head_init(&sk->receive_queue);
skb_queue_head_init(&sk->write_queue);
skb_queue_head_init(&sk->error_queue);
init_timer(&sk->timer);
sk->allocation = GFP_KERNEL;
sk->rcvbuf = sysctl_rmem_default;
sk->sndbuf = sysctl_wmem_default;
sk->state = TCP_CLOSE;
sk->zapped = 1;
sk->socket = sock;
1132 if(sock)
{
sk->type = sock->type;
sk->sleep = &sock->wait;
sock->sk = sk;
1137 } else
sk->sleep = NULL;
sk->dst_lock = RW_LOCK_UNLOCKED;
sk->callback_lock = RW_LOCK_UNLOCKED;
sk->state_change = sock_def_wakeup;
sk->data_ready = sock_def_readable;
sk->write_space = sock_def_write_space;
sk->error_report = sock_def_error_report;
sk->destruct = sock_def_destruct;
sk->peercred.pid = 0;
sk->peercred.uid = -1;
sk->peercred.gid = -1;
sk->rcvlowat = 1;
sk->rcvtimeo = MAX_SCHEDULE_TIMEOUT;
sk->sndtimeo = MAX_SCHEDULE_TIMEOUT;
atomic_set(&sk->refcnt, 1);
}