2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* 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.
|
|
|
|
*
|
|
|
|
*
|
2005-05-06 01:16:16 +02:00
|
|
|
* Authors: Ross Biro
|
2005-04-17 00:20:36 +02:00
|
|
|
* 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
|
2007-02-09 15:24:36 +01:00
|
|
|
* code. The ACK stuff can wait and needs major
|
2005-04-17 00:20:36 +02:00
|
|
|
* 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
|
|
|
|
* Chris Evans : Security fixes - signedness again
|
|
|
|
* Arnaldo C. Melo : cleanups, use skb_queue_purge
|
|
|
|
*
|
|
|
|
* 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.
|
|
|
|
*/
|
|
|
|
|
2006-01-11 21:17:47 +01:00
|
|
|
#include <linux/capability.h>
|
2005-04-17 00:20:36 +02:00
|
|
|
#include <linux/errno.h>
|
|
|
|
#include <linux/types.h>
|
|
|
|
#include <linux/socket.h>
|
|
|
|
#include <linux/in.h>
|
|
|
|
#include <linux/kernel.h>
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/proc_fs.h>
|
|
|
|
#include <linux/seq_file.h>
|
|
|
|
#include <linux/sched.h>
|
|
|
|
#include <linux/timer.h>
|
|
|
|
#include <linux/string.h>
|
|
|
|
#include <linux/sockios.h>
|
|
|
|
#include <linux/net.h>
|
|
|
|
#include <linux/mm.h>
|
|
|
|
#include <linux/slab.h>
|
|
|
|
#include <linux/interrupt.h>
|
|
|
|
#include <linux/poll.h>
|
|
|
|
#include <linux/tcp.h>
|
|
|
|
#include <linux/init.h>
|
2006-10-19 22:08:53 +02:00
|
|
|
#include <linux/highmem.h>
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
#include <asm/uaccess.h>
|
|
|
|
#include <asm/system.h>
|
|
|
|
|
|
|
|
#include <linux/netdevice.h>
|
|
|
|
#include <net/protocol.h>
|
|
|
|
#include <linux/skbuff.h>
|
2007-09-12 12:01:34 +02:00
|
|
|
#include <net/net_namespace.h>
|
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 07:46:52 +02:00
|
|
|
#include <net/request_sock.h>
|
2005-04-17 00:20:36 +02:00
|
|
|
#include <net/sock.h>
|
2009-02-12 06:03:38 +01:00
|
|
|
#include <linux/net_tstamp.h>
|
2005-04-17 00:20:36 +02:00
|
|
|
#include <net/xfrm.h>
|
|
|
|
#include <linux/ipsec.h>
|
|
|
|
|
|
|
|
#include <linux/filter.h>
|
|
|
|
|
|
|
|
#ifdef CONFIG_INET
|
|
|
|
#include <net/tcp.h>
|
|
|
|
#endif
|
|
|
|
|
2006-07-03 09:25:12 +02:00
|
|
|
/*
|
|
|
|
* Each address family might have different locking rules, so we have
|
|
|
|
* one slock key per address family:
|
|
|
|
*/
|
2006-07-03 09:25:35 +02:00
|
|
|
static struct lock_class_key af_family_keys[AF_MAX];
|
|
|
|
static struct lock_class_key af_family_slock_keys[AF_MAX];
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Make lock validator output more readable. (we pre-construct these
|
|
|
|
* strings build-time, so that runtime initialization of socket
|
|
|
|
* locks is fast):
|
|
|
|
*/
|
|
|
|
static const char *af_family_key_strings[AF_MAX+1] = {
|
|
|
|
"sk_lock-AF_UNSPEC", "sk_lock-AF_UNIX" , "sk_lock-AF_INET" ,
|
|
|
|
"sk_lock-AF_AX25" , "sk_lock-AF_IPX" , "sk_lock-AF_APPLETALK",
|
|
|
|
"sk_lock-AF_NETROM", "sk_lock-AF_BRIDGE" , "sk_lock-AF_ATMPVC" ,
|
|
|
|
"sk_lock-AF_X25" , "sk_lock-AF_INET6" , "sk_lock-AF_ROSE" ,
|
|
|
|
"sk_lock-AF_DECnet", "sk_lock-AF_NETBEUI" , "sk_lock-AF_SECURITY" ,
|
|
|
|
"sk_lock-AF_KEY" , "sk_lock-AF_NETLINK" , "sk_lock-AF_PACKET" ,
|
|
|
|
"sk_lock-AF_ASH" , "sk_lock-AF_ECONET" , "sk_lock-AF_ATMSVC" ,
|
2009-02-27 08:43:19 +01:00
|
|
|
"sk_lock-AF_RDS" , "sk_lock-AF_SNA" , "sk_lock-AF_IRDA" ,
|
2006-07-03 09:25:35 +02:00
|
|
|
"sk_lock-AF_PPPOX" , "sk_lock-AF_WANPIPE" , "sk_lock-AF_LLC" ,
|
2007-12-17 00:59:24 +01:00
|
|
|
"sk_lock-27" , "sk_lock-28" , "sk_lock-AF_CAN" ,
|
2007-04-27 00:48:28 +02:00
|
|
|
"sk_lock-AF_TIPC" , "sk_lock-AF_BLUETOOTH", "sk_lock-IUCV" ,
|
2008-09-23 04:51:15 +02:00
|
|
|
"sk_lock-AF_RXRPC" , "sk_lock-AF_ISDN" , "sk_lock-AF_PHONET" ,
|
2009-06-08 14:18:47 +02:00
|
|
|
"sk_lock-AF_IEEE802154",
|
2008-09-23 04:51:15 +02:00
|
|
|
"sk_lock-AF_MAX"
|
2006-07-03 09:25:35 +02:00
|
|
|
};
|
|
|
|
static const char *af_family_slock_key_strings[AF_MAX+1] = {
|
|
|
|
"slock-AF_UNSPEC", "slock-AF_UNIX" , "slock-AF_INET" ,
|
|
|
|
"slock-AF_AX25" , "slock-AF_IPX" , "slock-AF_APPLETALK",
|
|
|
|
"slock-AF_NETROM", "slock-AF_BRIDGE" , "slock-AF_ATMPVC" ,
|
|
|
|
"slock-AF_X25" , "slock-AF_INET6" , "slock-AF_ROSE" ,
|
|
|
|
"slock-AF_DECnet", "slock-AF_NETBEUI" , "slock-AF_SECURITY" ,
|
|
|
|
"slock-AF_KEY" , "slock-AF_NETLINK" , "slock-AF_PACKET" ,
|
|
|
|
"slock-AF_ASH" , "slock-AF_ECONET" , "slock-AF_ATMSVC" ,
|
2009-02-27 08:43:19 +01:00
|
|
|
"slock-AF_RDS" , "slock-AF_SNA" , "slock-AF_IRDA" ,
|
2006-07-03 09:25:35 +02:00
|
|
|
"slock-AF_PPPOX" , "slock-AF_WANPIPE" , "slock-AF_LLC" ,
|
2007-12-17 00:59:24 +01:00
|
|
|
"slock-27" , "slock-28" , "slock-AF_CAN" ,
|
2007-04-27 00:48:28 +02:00
|
|
|
"slock-AF_TIPC" , "slock-AF_BLUETOOTH", "slock-AF_IUCV" ,
|
2008-09-23 04:51:15 +02:00
|
|
|
"slock-AF_RXRPC" , "slock-AF_ISDN" , "slock-AF_PHONET" ,
|
2009-06-08 14:18:47 +02:00
|
|
|
"slock-AF_IEEE802154",
|
2008-09-23 04:51:15 +02:00
|
|
|
"slock-AF_MAX"
|
2006-07-03 09:25:35 +02:00
|
|
|
};
|
2007-07-19 10:49:00 +02:00
|
|
|
static const char *af_family_clock_key_strings[AF_MAX+1] = {
|
|
|
|
"clock-AF_UNSPEC", "clock-AF_UNIX" , "clock-AF_INET" ,
|
|
|
|
"clock-AF_AX25" , "clock-AF_IPX" , "clock-AF_APPLETALK",
|
|
|
|
"clock-AF_NETROM", "clock-AF_BRIDGE" , "clock-AF_ATMPVC" ,
|
|
|
|
"clock-AF_X25" , "clock-AF_INET6" , "clock-AF_ROSE" ,
|
|
|
|
"clock-AF_DECnet", "clock-AF_NETBEUI" , "clock-AF_SECURITY" ,
|
|
|
|
"clock-AF_KEY" , "clock-AF_NETLINK" , "clock-AF_PACKET" ,
|
|
|
|
"clock-AF_ASH" , "clock-AF_ECONET" , "clock-AF_ATMSVC" ,
|
2009-02-27 08:43:19 +01:00
|
|
|
"clock-AF_RDS" , "clock-AF_SNA" , "clock-AF_IRDA" ,
|
2007-07-19 10:49:00 +02:00
|
|
|
"clock-AF_PPPOX" , "clock-AF_WANPIPE" , "clock-AF_LLC" ,
|
2008-07-23 23:06:04 +02:00
|
|
|
"clock-27" , "clock-28" , "clock-AF_CAN" ,
|
2007-07-22 04:30:16 +02:00
|
|
|
"clock-AF_TIPC" , "clock-AF_BLUETOOTH", "clock-AF_IUCV" ,
|
2008-09-23 04:51:15 +02:00
|
|
|
"clock-AF_RXRPC" , "clock-AF_ISDN" , "clock-AF_PHONET" ,
|
2009-06-08 14:18:47 +02:00
|
|
|
"clock-AF_IEEE802154",
|
2008-09-23 04:51:15 +02:00
|
|
|
"clock-AF_MAX"
|
2007-07-19 10:49:00 +02:00
|
|
|
};
|
2006-07-03 09:25:12 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* sk_callback_lock locking rules are per-address-family,
|
|
|
|
* so split the lock classes by using a per-AF key:
|
|
|
|
*/
|
|
|
|
static struct lock_class_key af_callback_keys[AF_MAX];
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/* Take into consideration the size of the struct sk_buff overhead in the
|
|
|
|
* determination of these values, since that is non-constant across
|
|
|
|
* platforms. This makes socket queueing behavior and performance
|
|
|
|
* not depend upon such differences.
|
|
|
|
*/
|
|
|
|
#define _SK_MEM_PACKETS 256
|
|
|
|
#define _SK_MEM_OVERHEAD (sizeof(struct sk_buff) + 256)
|
|
|
|
#define SK_WMEM_MAX (_SK_MEM_OVERHEAD * _SK_MEM_PACKETS)
|
|
|
|
#define SK_RMEM_MAX (_SK_MEM_OVERHEAD * _SK_MEM_PACKETS)
|
|
|
|
|
|
|
|
/* Run time adjustable parameters. */
|
2006-09-22 23:15:41 +02:00
|
|
|
__u32 sysctl_wmem_max __read_mostly = SK_WMEM_MAX;
|
|
|
|
__u32 sysctl_rmem_max __read_mostly = SK_RMEM_MAX;
|
|
|
|
__u32 sysctl_wmem_default __read_mostly = SK_WMEM_MAX;
|
|
|
|
__u32 sysctl_rmem_default __read_mostly = SK_RMEM_MAX;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/* Maximal space eaten by iovec or ancilliary data plus some space */
|
2006-09-22 23:15:41 +02:00
|
|
|
int sysctl_optmem_max __read_mostly = sizeof(unsigned long)*(2*UIO_MAXIOV+512);
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sysctl_optmem_max);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
static int sock_set_timeout(long *timeo_p, char __user *optval, int optlen)
|
|
|
|
{
|
|
|
|
struct timeval tv;
|
|
|
|
|
|
|
|
if (optlen < sizeof(tv))
|
|
|
|
return -EINVAL;
|
|
|
|
if (copy_from_user(&tv, optval, sizeof(tv)))
|
|
|
|
return -EFAULT;
|
2007-05-25 01:58:54 +02:00
|
|
|
if (tv.tv_usec < 0 || tv.tv_usec >= USEC_PER_SEC)
|
|
|
|
return -EDOM;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-05-25 01:58:54 +02:00
|
|
|
if (tv.tv_sec < 0) {
|
2007-07-09 22:16:00 +02:00
|
|
|
static int warned __read_mostly;
|
|
|
|
|
2007-05-25 01:58:54 +02:00
|
|
|
*timeo_p = 0;
|
2008-05-03 01:20:10 +02:00
|
|
|
if (warned < 10 && net_ratelimit()) {
|
2007-05-25 01:58:54 +02:00
|
|
|
warned++;
|
|
|
|
printk(KERN_INFO "sock_set_timeout: `%s' (pid %d) "
|
|
|
|
"tries to set negative timeout\n",
|
2007-10-19 08:40:40 +02:00
|
|
|
current->comm, task_pid_nr(current));
|
2008-05-03 01:20:10 +02:00
|
|
|
}
|
2007-05-25 01:58:54 +02:00
|
|
|
return 0;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
*timeo_p = MAX_SCHEDULE_TIMEOUT;
|
|
|
|
if (tv.tv_sec == 0 && tv.tv_usec == 0)
|
|
|
|
return 0;
|
|
|
|
if (tv.tv_sec < (MAX_SCHEDULE_TIMEOUT/HZ - 1))
|
|
|
|
*timeo_p = tv.tv_sec*HZ + (tv.tv_usec+(1000000/HZ-1))/(1000000/HZ);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void sock_warn_obsolete_bsdism(const char *name)
|
|
|
|
{
|
|
|
|
static int warned;
|
|
|
|
static char warncomm[TASK_COMM_LEN];
|
2007-02-09 15:24:36 +01:00
|
|
|
if (strcmp(warncomm, current->comm) && warned < 5) {
|
|
|
|
strcpy(warncomm, current->comm);
|
2005-04-17 00:20:36 +02:00
|
|
|
printk(KERN_WARNING "process `%s' is using obsolete "
|
|
|
|
"%s SO_BSDCOMPAT\n", warncomm, name);
|
|
|
|
warned++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-02-12 06:03:38 +01:00
|
|
|
static void sock_disable_timestamp(struct sock *sk, int flag)
|
2007-02-09 15:24:36 +01:00
|
|
|
{
|
2009-02-12 06:03:38 +01:00
|
|
|
if (sock_flag(sk, flag)) {
|
|
|
|
sock_reset_flag(sk, flag);
|
|
|
|
if (!sock_flag(sk, SOCK_TIMESTAMP) &&
|
|
|
|
!sock_flag(sk, SOCK_TIMESTAMPING_RX_SOFTWARE)) {
|
|
|
|
net_disable_timestamp();
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2006-03-28 11:08:21 +02:00
|
|
|
int sock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
int err = 0;
|
|
|
|
int skb_len;
|
|
|
|
|
2008-05-14 12:50:03 +02:00
|
|
|
/* Cast sk->rcvbuf to unsigned... It's pointless, but reduces
|
2006-03-28 11:08:21 +02:00
|
|
|
number of warnings when compiling with -W --ANK
|
|
|
|
*/
|
|
|
|
if (atomic_read(&sk->sk_rmem_alloc) + skb->truesize >=
|
|
|
|
(unsigned)sk->sk_rcvbuf) {
|
|
|
|
err = -ENOMEM;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2006-09-01 00:28:39 +02:00
|
|
|
err = sk_filter(sk, skb);
|
2006-03-28 11:08:21 +02:00
|
|
|
if (err)
|
|
|
|
goto out;
|
|
|
|
|
2007-12-31 09:11:19 +01:00
|
|
|
if (!sk_rmem_schedule(sk, skb->truesize)) {
|
|
|
|
err = -ENOBUFS;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2006-03-28 11:08:21 +02:00
|
|
|
skb->dev = NULL;
|
|
|
|
skb_set_owner_r(skb, sk);
|
2008-12-18 07:11:38 +01:00
|
|
|
|
2006-03-28 11:08:21 +02:00
|
|
|
/* Cache the SKB length before we tack it onto the receive
|
|
|
|
* queue. Once it is added it no longer belongs to us and
|
|
|
|
* may be freed by other threads of control pulling packets
|
|
|
|
* from the queue.
|
|
|
|
*/
|
|
|
|
skb_len = skb->len;
|
|
|
|
|
|
|
|
skb_queue_tail(&sk->sk_receive_queue, skb);
|
|
|
|
|
|
|
|
if (!sock_flag(sk, SOCK_DEAD))
|
|
|
|
sk->sk_data_ready(sk, skb_len);
|
|
|
|
out:
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(sock_queue_rcv_skb);
|
|
|
|
|
2006-11-16 17:06:06 +01:00
|
|
|
int sk_receive_skb(struct sock *sk, struct sk_buff *skb, const int nested)
|
2006-03-28 11:08:21 +02:00
|
|
|
{
|
|
|
|
int rc = NET_RX_SUCCESS;
|
|
|
|
|
2006-09-01 00:28:39 +02:00
|
|
|
if (sk_filter(sk, skb))
|
2006-03-28 11:08:21 +02:00
|
|
|
goto discard_and_relse;
|
|
|
|
|
|
|
|
skb->dev = NULL;
|
|
|
|
|
2006-11-16 17:06:06 +01:00
|
|
|
if (nested)
|
|
|
|
bh_lock_sock_nested(sk);
|
|
|
|
else
|
|
|
|
bh_lock_sock(sk);
|
2006-07-03 09:25:35 +02:00
|
|
|
if (!sock_owned_by_user(sk)) {
|
|
|
|
/*
|
|
|
|
* trylock + unlock semantics:
|
|
|
|
*/
|
|
|
|
mutex_acquire(&sk->sk_lock.dep_map, 0, 1, _RET_IP_);
|
|
|
|
|
2008-10-07 23:18:42 +02:00
|
|
|
rc = sk_backlog_rcv(sk, skb);
|
2006-07-03 09:25:35 +02:00
|
|
|
|
|
|
|
mutex_release(&sk->sk_lock.dep_map, 1, _RET_IP_);
|
|
|
|
} else
|
2006-03-28 11:08:21 +02:00
|
|
|
sk_add_backlog(sk, skb);
|
|
|
|
bh_unlock_sock(sk);
|
|
|
|
out:
|
|
|
|
sock_put(sk);
|
|
|
|
return rc;
|
|
|
|
discard_and_relse:
|
|
|
|
kfree_skb(skb);
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(sk_receive_skb);
|
|
|
|
|
|
|
|
struct dst_entry *__sk_dst_check(struct sock *sk, u32 cookie)
|
|
|
|
{
|
|
|
|
struct dst_entry *dst = sk->sk_dst_cache;
|
|
|
|
|
|
|
|
if (dst && dst->obsolete && dst->ops->check(dst, cookie) == NULL) {
|
|
|
|
sk->sk_dst_cache = NULL;
|
|
|
|
dst_release(dst);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
return dst;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(__sk_dst_check);
|
|
|
|
|
|
|
|
struct dst_entry *sk_dst_check(struct sock *sk, u32 cookie)
|
|
|
|
{
|
|
|
|
struct dst_entry *dst = sk_dst_get(sk);
|
|
|
|
|
|
|
|
if (dst && dst->obsolete && dst->ops->check(dst, cookie) == NULL) {
|
|
|
|
sk_dst_reset(sk);
|
|
|
|
dst_release(dst);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
return dst;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(sk_dst_check);
|
|
|
|
|
2007-09-15 01:41:03 +02:00
|
|
|
static int sock_bindtodevice(struct sock *sk, char __user *optval, int optlen)
|
|
|
|
{
|
|
|
|
int ret = -ENOPROTOOPT;
|
|
|
|
#ifdef CONFIG_NETDEVICES
|
2008-03-25 18:26:21 +01:00
|
|
|
struct net *net = sock_net(sk);
|
2007-09-15 01:41:03 +02:00
|
|
|
char devname[IFNAMSIZ];
|
|
|
|
int index;
|
|
|
|
|
|
|
|
/* Sorry... */
|
|
|
|
ret = -EPERM;
|
|
|
|
if (!capable(CAP_NET_RAW))
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
ret = -EINVAL;
|
|
|
|
if (optlen < 0)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
/* 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.
|
|
|
|
*/
|
|
|
|
if (optlen > IFNAMSIZ - 1)
|
|
|
|
optlen = IFNAMSIZ - 1;
|
|
|
|
memset(devname, 0, sizeof(devname));
|
|
|
|
|
|
|
|
ret = -EFAULT;
|
|
|
|
if (copy_from_user(devname, optval, optlen))
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
if (devname[0] == '\0') {
|
|
|
|
index = 0;
|
|
|
|
} else {
|
2007-09-17 20:56:21 +02:00
|
|
|
struct net_device *dev = dev_get_by_name(net, devname);
|
2007-09-15 01:41:03 +02:00
|
|
|
|
|
|
|
ret = -ENODEV;
|
|
|
|
if (!dev)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
index = dev->ifindex;
|
|
|
|
dev_put(dev);
|
|
|
|
}
|
|
|
|
|
|
|
|
lock_sock(sk);
|
|
|
|
sk->sk_bound_dev_if = index;
|
|
|
|
sk_dst_reset(sk);
|
|
|
|
release_sock(sk);
|
|
|
|
|
|
|
|
ret = 0;
|
|
|
|
|
|
|
|
out:
|
|
|
|
#endif
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2007-11-15 12:03:19 +01:00
|
|
|
static inline void sock_valbool_flag(struct sock *sk, int bit, int valbool)
|
|
|
|
{
|
|
|
|
if (valbool)
|
|
|
|
sock_set_flag(sk, bit);
|
|
|
|
else
|
|
|
|
sock_reset_flag(sk, bit);
|
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* This is meant for all protocols to use and covers goings on
|
|
|
|
* at the socket level. Everything here is generic.
|
|
|
|
*/
|
|
|
|
|
|
|
|
int sock_setsockopt(struct socket *sock, int level, int optname,
|
|
|
|
char __user *optval, int optlen)
|
|
|
|
{
|
2009-05-27 13:30:05 +02:00
|
|
|
struct sock *sk = sock->sk;
|
2005-04-17 00:20:36 +02:00
|
|
|
int val;
|
|
|
|
int valbool;
|
|
|
|
struct linger ling;
|
|
|
|
int ret = 0;
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* Options without arguments
|
|
|
|
*/
|
|
|
|
|
2007-09-15 01:41:03 +02:00
|
|
|
if (optname == SO_BINDTODEVICE)
|
|
|
|
return sock_bindtodevice(sk, optval, optlen);
|
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
if (optlen < sizeof(int))
|
|
|
|
return -EINVAL;
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
if (get_user(val, (int __user *)optval))
|
|
|
|
return -EFAULT;
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2009-05-27 13:30:05 +02:00
|
|
|
valbool = val ? 1 : 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
lock_sock(sk);
|
|
|
|
|
2009-05-27 13:30:05 +02:00
|
|
|
switch (optname) {
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_DEBUG:
|
2009-05-27 13:30:05 +02:00
|
|
|
if (val && !capable(CAP_NET_ADMIN))
|
2007-04-11 05:10:33 +02:00
|
|
|
ret = -EACCES;
|
2009-05-27 13:30:05 +02:00
|
|
|
else
|
2007-11-15 12:03:19 +01:00
|
|
|
sock_valbool_flag(sk, SOCK_DBG, valbool);
|
2007-04-11 05:10:33 +02:00
|
|
|
break;
|
|
|
|
case SO_REUSEADDR:
|
|
|
|
sk->sk_reuse = valbool;
|
|
|
|
break;
|
|
|
|
case SO_TYPE:
|
|
|
|
case SO_ERROR:
|
|
|
|
ret = -ENOPROTOOPT;
|
|
|
|
break;
|
|
|
|
case SO_DONTROUTE:
|
2007-11-15 12:03:19 +01:00
|
|
|
sock_valbool_flag(sk, SOCK_LOCALROUTE, valbool);
|
2007-04-11 05:10:33 +02:00
|
|
|
break;
|
|
|
|
case SO_BROADCAST:
|
|
|
|
sock_valbool_flag(sk, SOCK_BROADCAST, valbool);
|
|
|
|
break;
|
|
|
|
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 */
|
|
|
|
|
|
|
|
if (val > sysctl_wmem_max)
|
|
|
|
val = sysctl_wmem_max;
|
2005-08-10 04:30:51 +02:00
|
|
|
set_sndbuf:
|
2007-04-11 05:10:33 +02:00
|
|
|
sk->sk_userlocks |= SOCK_SNDBUF_LOCK;
|
|
|
|
if ((val * 2) < SOCK_MIN_SNDBUF)
|
|
|
|
sk->sk_sndbuf = SOCK_MIN_SNDBUF;
|
|
|
|
else
|
|
|
|
sk->sk_sndbuf = val * 2;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
/*
|
|
|
|
* Wake up sending tasks if we
|
|
|
|
* upped the value.
|
|
|
|
*/
|
|
|
|
sk->sk_write_space(sk);
|
|
|
|
break;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_SNDBUFFORCE:
|
|
|
|
if (!capable(CAP_NET_ADMIN)) {
|
|
|
|
ret = -EPERM;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
goto set_sndbuf;
|
2005-08-10 04:30:51 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
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 */
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
if (val > sysctl_rmem_max)
|
|
|
|
val = sysctl_rmem_max;
|
2005-08-10 04:30:51 +02:00
|
|
|
set_rcvbuf:
|
2007-04-11 05:10:33 +02:00
|
|
|
sk->sk_userlocks |= SOCK_RCVBUF_LOCK;
|
|
|
|
/*
|
|
|
|
* We double it on the way in to account for
|
|
|
|
* "struct sk_buff" etc. overhead. Applications
|
|
|
|
* assume that the SO_RCVBUF setting they make will
|
|
|
|
* allow that much actual data to be received on that
|
|
|
|
* socket.
|
|
|
|
*
|
|
|
|
* Applications are unaware that "struct sk_buff" and
|
|
|
|
* other overheads allocate from the receive buffer
|
|
|
|
* during socket buffer allocation.
|
|
|
|
*
|
|
|
|
* And after considering the possible alternatives,
|
|
|
|
* returning the value we actually used in getsockopt
|
|
|
|
* is the most desirable behavior.
|
|
|
|
*/
|
|
|
|
if ((val * 2) < SOCK_MIN_RCVBUF)
|
|
|
|
sk->sk_rcvbuf = SOCK_MIN_RCVBUF;
|
|
|
|
else
|
|
|
|
sk->sk_rcvbuf = val * 2;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_RCVBUFFORCE:
|
|
|
|
if (!capable(CAP_NET_ADMIN)) {
|
|
|
|
ret = -EPERM;
|
2005-04-17 00:20:36 +02:00
|
|
|
break;
|
2007-04-11 05:10:33 +02:00
|
|
|
}
|
|
|
|
goto set_rcvbuf;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_KEEPALIVE:
|
2005-04-17 00:20:36 +02:00
|
|
|
#ifdef CONFIG_INET
|
2007-04-11 05:10:33 +02:00
|
|
|
if (sk->sk_protocol == IPPROTO_TCP)
|
|
|
|
tcp_set_keepalive(sk, valbool);
|
2005-04-17 00:20:36 +02:00
|
|
|
#endif
|
2007-04-11 05:10:33 +02:00
|
|
|
sock_valbool_flag(sk, SOCK_KEEPOPEN, valbool);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_OOBINLINE:
|
|
|
|
sock_valbool_flag(sk, SOCK_URGINLINE, valbool);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_NO_CHECK:
|
|
|
|
sk->sk_no_check = valbool;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_PRIORITY:
|
|
|
|
if ((val >= 0 && val <= 6) || capable(CAP_NET_ADMIN))
|
|
|
|
sk->sk_priority = val;
|
|
|
|
else
|
|
|
|
ret = -EPERM;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_LINGER:
|
|
|
|
if (optlen < sizeof(ling)) {
|
|
|
|
ret = -EINVAL; /* 1003.1g */
|
2005-04-17 00:20:36 +02:00
|
|
|
break;
|
2007-04-11 05:10:33 +02:00
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
if (copy_from_user(&ling, optval, sizeof(ling))) {
|
2007-04-11 05:10:33 +02:00
|
|
|
ret = -EFAULT;
|
2005-04-17 00:20:36 +02:00
|
|
|
break;
|
2007-04-11 05:10:33 +02:00
|
|
|
}
|
|
|
|
if (!ling.l_onoff)
|
|
|
|
sock_reset_flag(sk, SOCK_LINGER);
|
|
|
|
else {
|
2005-04-17 00:20:36 +02:00
|
|
|
#if (BITS_PER_LONG == 32)
|
2007-04-11 05:10:33 +02:00
|
|
|
if ((unsigned int)ling.l_linger >= MAX_SCHEDULE_TIMEOUT/HZ)
|
|
|
|
sk->sk_lingertime = MAX_SCHEDULE_TIMEOUT;
|
2005-04-17 00:20:36 +02:00
|
|
|
else
|
2007-04-11 05:10:33 +02:00
|
|
|
#endif
|
|
|
|
sk->sk_lingertime = (unsigned int)ling.l_linger * HZ;
|
|
|
|
sock_set_flag(sk, SOCK_LINGER);
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_BSDCOMPAT:
|
|
|
|
sock_warn_obsolete_bsdism("setsockopt");
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_PASSCRED:
|
|
|
|
if (valbool)
|
|
|
|
set_bit(SOCK_PASSCRED, &sock->flags);
|
|
|
|
else
|
|
|
|
clear_bit(SOCK_PASSCRED, &sock->flags);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_TIMESTAMP:
|
2007-03-26 07:14:49 +02:00
|
|
|
case SO_TIMESTAMPNS:
|
2007-04-11 05:10:33 +02:00
|
|
|
if (valbool) {
|
2007-03-26 07:14:49 +02:00
|
|
|
if (optname == SO_TIMESTAMP)
|
|
|
|
sock_reset_flag(sk, SOCK_RCVTSTAMPNS);
|
|
|
|
else
|
|
|
|
sock_set_flag(sk, SOCK_RCVTSTAMPNS);
|
2007-04-11 05:10:33 +02:00
|
|
|
sock_set_flag(sk, SOCK_RCVTSTAMP);
|
2009-02-12 06:03:38 +01:00
|
|
|
sock_enable_timestamp(sk, SOCK_TIMESTAMP);
|
2007-03-26 07:14:49 +02:00
|
|
|
} else {
|
2007-04-11 05:10:33 +02:00
|
|
|
sock_reset_flag(sk, SOCK_RCVTSTAMP);
|
2007-03-26 07:14:49 +02:00
|
|
|
sock_reset_flag(sk, SOCK_RCVTSTAMPNS);
|
|
|
|
}
|
2007-04-11 05:10:33 +02:00
|
|
|
break;
|
|
|
|
|
2009-02-12 06:03:38 +01:00
|
|
|
case SO_TIMESTAMPING:
|
|
|
|
if (val & ~SOF_TIMESTAMPING_MASK) {
|
2009-07-20 02:47:04 +02:00
|
|
|
ret = -EINVAL;
|
2009-02-12 06:03:38 +01:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
sock_valbool_flag(sk, SOCK_TIMESTAMPING_TX_HARDWARE,
|
|
|
|
val & SOF_TIMESTAMPING_TX_HARDWARE);
|
|
|
|
sock_valbool_flag(sk, SOCK_TIMESTAMPING_TX_SOFTWARE,
|
|
|
|
val & SOF_TIMESTAMPING_TX_SOFTWARE);
|
|
|
|
sock_valbool_flag(sk, SOCK_TIMESTAMPING_RX_HARDWARE,
|
|
|
|
val & SOF_TIMESTAMPING_RX_HARDWARE);
|
|
|
|
if (val & SOF_TIMESTAMPING_RX_SOFTWARE)
|
|
|
|
sock_enable_timestamp(sk,
|
|
|
|
SOCK_TIMESTAMPING_RX_SOFTWARE);
|
|
|
|
else
|
|
|
|
sock_disable_timestamp(sk,
|
|
|
|
SOCK_TIMESTAMPING_RX_SOFTWARE);
|
|
|
|
sock_valbool_flag(sk, SOCK_TIMESTAMPING_SOFTWARE,
|
|
|
|
val & SOF_TIMESTAMPING_SOFTWARE);
|
|
|
|
sock_valbool_flag(sk, SOCK_TIMESTAMPING_SYS_HARDWARE,
|
|
|
|
val & SOF_TIMESTAMPING_SYS_HARDWARE);
|
|
|
|
sock_valbool_flag(sk, SOCK_TIMESTAMPING_RAW_HARDWARE,
|
|
|
|
val & SOF_TIMESTAMPING_RAW_HARDWARE);
|
|
|
|
break;
|
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_RCVLOWAT:
|
|
|
|
if (val < 0)
|
|
|
|
val = INT_MAX;
|
|
|
|
sk->sk_rcvlowat = val ? : 1;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_RCVTIMEO:
|
|
|
|
ret = sock_set_timeout(&sk->sk_rcvtimeo, optval, optlen);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_SNDTIMEO:
|
|
|
|
ret = sock_set_timeout(&sk->sk_sndtimeo, optval, optlen);
|
|
|
|
break;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_ATTACH_FILTER:
|
|
|
|
ret = -EINVAL;
|
|
|
|
if (optlen == sizeof(struct sock_fprog)) {
|
|
|
|
struct sock_fprog fprog;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
ret = -EFAULT;
|
|
|
|
if (copy_from_user(&fprog, optval, sizeof(fprog)))
|
2005-04-17 00:20:36 +02:00
|
|
|
break;
|
2007-04-11 05:10:33 +02:00
|
|
|
|
|
|
|
ret = sk_attach_filter(&fprog, sk);
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_DETACH_FILTER:
|
2007-10-18 06:21:26 +02:00
|
|
|
ret = sk_detach_filter(sk);
|
2007-04-11 05:10:33 +02:00
|
|
|
break;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_PASSSEC:
|
|
|
|
if (valbool)
|
|
|
|
set_bit(SOCK_PASSSEC, &sock->flags);
|
|
|
|
else
|
|
|
|
clear_bit(SOCK_PASSSEC, &sock->flags);
|
|
|
|
break;
|
2008-01-31 04:08:16 +01:00
|
|
|
case SO_MARK:
|
|
|
|
if (!capable(CAP_NET_ADMIN))
|
|
|
|
ret = -EPERM;
|
2009-05-27 13:30:05 +02:00
|
|
|
else
|
2008-01-31 04:08:16 +01:00
|
|
|
sk->sk_mark = val;
|
|
|
|
break;
|
[AF_UNIX]: Datagram getpeersec
This patch implements an API whereby an application can determine the
label of its peer's Unix datagram sockets via the auxiliary data mechanism of
recvmsg.
Patch purpose:
This patch enables a security-aware application to retrieve the
security context of the peer of a Unix datagram socket. The application
can then use this security context to determine the security context for
processing on behalf of the peer who sent the packet.
Patch design and implementation:
The design and implementation is very similar to the UDP case for INET
sockets. Basically we build upon the existing Unix domain socket API for
retrieving user credentials. Linux offers the API for obtaining user
credentials via ancillary messages (i.e., out of band/control messages
that are bundled together with a normal message). To retrieve the security
context, the application first indicates to the kernel such desire by
setting the SO_PASSSEC option via getsockopt. Then the application
retrieves the security context using the auxiliary data mechanism.
An example server application for Unix datagram socket should look like this:
toggle = 1;
toggle_len = sizeof(toggle);
setsockopt(sockfd, SOL_SOCKET, SO_PASSSEC, &toggle, &toggle_len);
recvmsg(sockfd, &msg_hdr, 0);
if (msg_hdr.msg_controllen > sizeof(struct cmsghdr)) {
cmsg_hdr = CMSG_FIRSTHDR(&msg_hdr);
if (cmsg_hdr->cmsg_len <= CMSG_LEN(sizeof(scontext)) &&
cmsg_hdr->cmsg_level == SOL_SOCKET &&
cmsg_hdr->cmsg_type == SCM_SECURITY) {
memcpy(&scontext, CMSG_DATA(cmsg_hdr), sizeof(scontext));
}
}
sock_setsockopt is enhanced with a new socket option SOCK_PASSSEC to allow
a server socket to receive security context of the peer.
Testing:
We have tested the patch by setting up Unix datagram client and server
applications. We verified that the server can retrieve the security context
using the auxiliary data mechanism of recvmsg.
Signed-off-by: Catherine Zhang <cxzhang@watson.ibm.com>
Acked-by: Acked-by: James Morris <jmorris@namei.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2006-06-29 21:27:47 +02:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/* We implement the SO_SNDLOWAT etc to
|
|
|
|
not be settable (1003.1g 5.3) */
|
2007-04-11 05:10:33 +02:00
|
|
|
default:
|
|
|
|
ret = -ENOPROTOOPT;
|
|
|
|
break;
|
2007-02-09 15:24:36 +01:00
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
release_sock(sk);
|
|
|
|
return ret;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_setsockopt);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
|
|
|
|
int sock_getsockopt(struct socket *sock, int level, int optname,
|
|
|
|
char __user *optval, int __user *optlen)
|
|
|
|
{
|
|
|
|
struct sock *sk = sock->sk;
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
union {
|
2007-02-09 15:24:36 +01:00
|
|
|
int val;
|
|
|
|
struct linger ling;
|
2005-04-17 00:20:36 +02:00
|
|
|
struct timeval tm;
|
|
|
|
} v;
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
unsigned int lv = sizeof(int);
|
|
|
|
int len;
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
if (get_user(len, optlen))
|
2007-02-09 15:24:36 +01:00
|
|
|
return -EFAULT;
|
2007-04-11 05:10:33 +02:00
|
|
|
if (len < 0)
|
2005-04-17 00:20:36 +02:00
|
|
|
return -EINVAL;
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2009-02-24 00:38:41 +01:00
|
|
|
memset(&v, 0, sizeof(v));
|
net: 4 bytes kernel memory disclosure in SO_BSDCOMPAT gsopt try #2
In function sock_getsockopt() located in net/core/sock.c, optval v.val
is not correctly initialized and directly returned in userland in case
we have SO_BSDCOMPAT option set.
This dummy code should trigger the bug:
int main(void)
{
unsigned char buf[4] = { 0, 0, 0, 0 };
int len;
int sock;
sock = socket(33, 2, 2);
getsockopt(sock, 1, SO_BSDCOMPAT, &buf, &len);
printf("%x%x%x%x\n", buf[0], buf[1], buf[2], buf[3]);
close(sock);
}
Here is a patch that fix this bug by initalizing v.val just after its
declaration.
Signed-off-by: Clément Lecigne <clement.lecigne@netasq.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-02-13 01:59:09 +01:00
|
|
|
|
2009-05-27 13:30:05 +02:00
|
|
|
switch (optname) {
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_DEBUG:
|
|
|
|
v.val = sock_flag(sk, SOCK_DBG);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_DONTROUTE:
|
|
|
|
v.val = sock_flag(sk, SOCK_LOCALROUTE);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_BROADCAST:
|
|
|
|
v.val = !!sock_flag(sk, SOCK_BROADCAST);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_SNDBUF:
|
|
|
|
v.val = sk->sk_sndbuf;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_RCVBUF:
|
|
|
|
v.val = sk->sk_rcvbuf;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_REUSEADDR:
|
|
|
|
v.val = sk->sk_reuse;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_KEEPALIVE:
|
|
|
|
v.val = !!sock_flag(sk, SOCK_KEEPOPEN);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_TYPE:
|
|
|
|
v.val = sk->sk_type;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_ERROR:
|
|
|
|
v.val = -sock_error(sk);
|
2009-05-27 13:30:05 +02:00
|
|
|
if (v.val == 0)
|
2007-04-11 05:10:33 +02:00
|
|
|
v.val = xchg(&sk->sk_err_soft, 0);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_OOBINLINE:
|
|
|
|
v.val = !!sock_flag(sk, SOCK_URGINLINE);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_NO_CHECK:
|
|
|
|
v.val = sk->sk_no_check;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_PRIORITY:
|
|
|
|
v.val = sk->sk_priority;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_LINGER:
|
|
|
|
lv = sizeof(v.ling);
|
|
|
|
v.ling.l_onoff = !!sock_flag(sk, SOCK_LINGER);
|
|
|
|
v.ling.l_linger = sk->sk_lingertime / HZ;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_BSDCOMPAT:
|
|
|
|
sock_warn_obsolete_bsdism("getsockopt");
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_TIMESTAMP:
|
2007-03-26 07:14:49 +02:00
|
|
|
v.val = sock_flag(sk, SOCK_RCVTSTAMP) &&
|
|
|
|
!sock_flag(sk, SOCK_RCVTSTAMPNS);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_TIMESTAMPNS:
|
|
|
|
v.val = sock_flag(sk, SOCK_RCVTSTAMPNS);
|
2007-04-11 05:10:33 +02:00
|
|
|
break;
|
|
|
|
|
2009-02-12 06:03:38 +01:00
|
|
|
case SO_TIMESTAMPING:
|
|
|
|
v.val = 0;
|
|
|
|
if (sock_flag(sk, SOCK_TIMESTAMPING_TX_HARDWARE))
|
|
|
|
v.val |= SOF_TIMESTAMPING_TX_HARDWARE;
|
|
|
|
if (sock_flag(sk, SOCK_TIMESTAMPING_TX_SOFTWARE))
|
|
|
|
v.val |= SOF_TIMESTAMPING_TX_SOFTWARE;
|
|
|
|
if (sock_flag(sk, SOCK_TIMESTAMPING_RX_HARDWARE))
|
|
|
|
v.val |= SOF_TIMESTAMPING_RX_HARDWARE;
|
|
|
|
if (sock_flag(sk, SOCK_TIMESTAMPING_RX_SOFTWARE))
|
|
|
|
v.val |= SOF_TIMESTAMPING_RX_SOFTWARE;
|
|
|
|
if (sock_flag(sk, SOCK_TIMESTAMPING_SOFTWARE))
|
|
|
|
v.val |= SOF_TIMESTAMPING_SOFTWARE;
|
|
|
|
if (sock_flag(sk, SOCK_TIMESTAMPING_SYS_HARDWARE))
|
|
|
|
v.val |= SOF_TIMESTAMPING_SYS_HARDWARE;
|
|
|
|
if (sock_flag(sk, SOCK_TIMESTAMPING_RAW_HARDWARE))
|
|
|
|
v.val |= SOF_TIMESTAMPING_RAW_HARDWARE;
|
|
|
|
break;
|
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_RCVTIMEO:
|
2009-05-27 13:30:05 +02:00
|
|
|
lv = sizeof(struct timeval);
|
2007-04-11 05:10:33 +02:00
|
|
|
if (sk->sk_rcvtimeo == MAX_SCHEDULE_TIMEOUT) {
|
|
|
|
v.tm.tv_sec = 0;
|
|
|
|
v.tm.tv_usec = 0;
|
|
|
|
} else {
|
|
|
|
v.tm.tv_sec = sk->sk_rcvtimeo / HZ;
|
|
|
|
v.tm.tv_usec = ((sk->sk_rcvtimeo % HZ) * 1000000) / HZ;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SO_SNDTIMEO:
|
2009-05-27 13:30:05 +02:00
|
|
|
lv = sizeof(struct timeval);
|
2007-04-11 05:10:33 +02:00
|
|
|
if (sk->sk_sndtimeo == MAX_SCHEDULE_TIMEOUT) {
|
|
|
|
v.tm.tv_sec = 0;
|
|
|
|
v.tm.tv_usec = 0;
|
|
|
|
} else {
|
|
|
|
v.tm.tv_sec = sk->sk_sndtimeo / HZ;
|
|
|
|
v.tm.tv_usec = ((sk->sk_sndtimeo % HZ) * 1000000) / HZ;
|
|
|
|
}
|
|
|
|
break;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_RCVLOWAT:
|
|
|
|
v.val = sk->sk_rcvlowat;
|
|
|
|
break;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_SNDLOWAT:
|
2009-05-27 13:30:05 +02:00
|
|
|
v.val = 1;
|
2007-04-11 05:10:33 +02:00
|
|
|
break;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_PASSCRED:
|
|
|
|
v.val = test_bit(SOCK_PASSCRED, &sock->flags) ? 1 : 0;
|
|
|
|
break;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_PEERCRED:
|
|
|
|
if (len > sizeof(sk->sk_peercred))
|
|
|
|
len = sizeof(sk->sk_peercred);
|
|
|
|
if (copy_to_user(optval, &sk->sk_peercred, len))
|
|
|
|
return -EFAULT;
|
|
|
|
goto lenout;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_PEERNAME:
|
|
|
|
{
|
|
|
|
char address[128];
|
|
|
|
|
|
|
|
if (sock->ops->getname(sock, (struct sockaddr *)address, &lv, 2))
|
|
|
|
return -ENOTCONN;
|
|
|
|
if (lv < len)
|
|
|
|
return -EINVAL;
|
|
|
|
if (copy_to_user(optval, address, len))
|
|
|
|
return -EFAULT;
|
|
|
|
goto lenout;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
/* Dubious BSD thing... Probably nobody even uses it, but
|
|
|
|
* the UNIX standard wants it for whatever reason... -DaveM
|
|
|
|
*/
|
|
|
|
case SO_ACCEPTCONN:
|
|
|
|
v.val = sk->sk_state == TCP_LISTEN;
|
|
|
|
break;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_PASSSEC:
|
|
|
|
v.val = test_bit(SOCK_PASSSEC, &sock->flags) ? 1 : 0;
|
|
|
|
break;
|
[AF_UNIX]: Datagram getpeersec
This patch implements an API whereby an application can determine the
label of its peer's Unix datagram sockets via the auxiliary data mechanism of
recvmsg.
Patch purpose:
This patch enables a security-aware application to retrieve the
security context of the peer of a Unix datagram socket. The application
can then use this security context to determine the security context for
processing on behalf of the peer who sent the packet.
Patch design and implementation:
The design and implementation is very similar to the UDP case for INET
sockets. Basically we build upon the existing Unix domain socket API for
retrieving user credentials. Linux offers the API for obtaining user
credentials via ancillary messages (i.e., out of band/control messages
that are bundled together with a normal message). To retrieve the security
context, the application first indicates to the kernel such desire by
setting the SO_PASSSEC option via getsockopt. Then the application
retrieves the security context using the auxiliary data mechanism.
An example server application for Unix datagram socket should look like this:
toggle = 1;
toggle_len = sizeof(toggle);
setsockopt(sockfd, SOL_SOCKET, SO_PASSSEC, &toggle, &toggle_len);
recvmsg(sockfd, &msg_hdr, 0);
if (msg_hdr.msg_controllen > sizeof(struct cmsghdr)) {
cmsg_hdr = CMSG_FIRSTHDR(&msg_hdr);
if (cmsg_hdr->cmsg_len <= CMSG_LEN(sizeof(scontext)) &&
cmsg_hdr->cmsg_level == SOL_SOCKET &&
cmsg_hdr->cmsg_type == SCM_SECURITY) {
memcpy(&scontext, CMSG_DATA(cmsg_hdr), sizeof(scontext));
}
}
sock_setsockopt is enhanced with a new socket option SOCK_PASSSEC to allow
a server socket to receive security context of the peer.
Testing:
We have tested the patch by setting up Unix datagram client and server
applications. We verified that the server can retrieve the security context
using the auxiliary data mechanism of recvmsg.
Signed-off-by: Catherine Zhang <cxzhang@watson.ibm.com>
Acked-by: Acked-by: James Morris <jmorris@namei.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2006-06-29 21:27:47 +02:00
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
case SO_PEERSEC:
|
|
|
|
return security_socket_getpeersec_stream(sock, optval, optlen, len);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2008-01-31 04:08:16 +01:00
|
|
|
case SO_MARK:
|
|
|
|
v.val = sk->sk_mark;
|
|
|
|
break;
|
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
default:
|
|
|
|
return -ENOPROTOOPT;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2007-04-11 05:10:33 +02:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
if (len > lv)
|
|
|
|
len = lv;
|
|
|
|
if (copy_to_user(optval, &v, len))
|
|
|
|
return -EFAULT;
|
|
|
|
lenout:
|
2007-02-09 15:24:36 +01:00
|
|
|
if (put_user(len, optlen))
|
|
|
|
return -EFAULT;
|
|
|
|
return 0;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2006-07-03 09:25:35 +02:00
|
|
|
/*
|
|
|
|
* Initialize an sk_lock.
|
|
|
|
*
|
|
|
|
* (We also register the sk_lock with the lock validator.)
|
|
|
|
*/
|
2007-03-22 20:27:49 +01:00
|
|
|
static inline void sock_lock_init(struct sock *sk)
|
2006-07-03 09:25:35 +02:00
|
|
|
{
|
2006-12-07 05:35:24 +01:00
|
|
|
sock_lock_init_class_and_name(sk,
|
|
|
|
af_family_slock_key_strings[sk->sk_family],
|
|
|
|
af_family_slock_keys + sk->sk_family,
|
|
|
|
af_family_key_strings[sk->sk_family],
|
|
|
|
af_family_keys + sk->sk_family);
|
2006-07-03 09:25:35 +02:00
|
|
|
}
|
|
|
|
|
2009-07-16 01:13:10 +02:00
|
|
|
/*
|
|
|
|
* Copy all fields from osk to nsk but nsk->sk_refcnt must not change yet,
|
|
|
|
* even temporarly, because of RCU lookups. sk_node should also be left as is.
|
|
|
|
*/
|
2007-11-01 08:29:45 +01:00
|
|
|
static void sock_copy(struct sock *nsk, const struct sock *osk)
|
|
|
|
{
|
|
|
|
#ifdef CONFIG_SECURITY_NETWORK
|
|
|
|
void *sptr = nsk->sk_security;
|
|
|
|
#endif
|
2009-07-16 01:13:10 +02:00
|
|
|
BUILD_BUG_ON(offsetof(struct sock, sk_copy_start) !=
|
|
|
|
sizeof(osk->sk_node) + sizeof(osk->sk_refcnt));
|
|
|
|
memcpy(&nsk->sk_copy_start, &osk->sk_copy_start,
|
|
|
|
osk->sk_prot->obj_size - offsetof(struct sock, sk_copy_start));
|
2007-11-01 08:29:45 +01:00
|
|
|
#ifdef CONFIG_SECURITY_NETWORK
|
|
|
|
nsk->sk_security = sptr;
|
|
|
|
security_sk_clone(osk, nsk);
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2007-11-01 08:36:26 +01:00
|
|
|
static struct sock *sk_prot_alloc(struct proto *prot, gfp_t priority,
|
|
|
|
int family)
|
2007-11-01 08:33:50 +01:00
|
|
|
{
|
|
|
|
struct sock *sk;
|
|
|
|
struct kmem_cache *slab;
|
|
|
|
|
|
|
|
slab = prot->slab;
|
2009-07-08 21:36:05 +02:00
|
|
|
if (slab != NULL) {
|
|
|
|
sk = kmem_cache_alloc(slab, priority & ~__GFP_ZERO);
|
|
|
|
if (!sk)
|
|
|
|
return sk;
|
|
|
|
if (priority & __GFP_ZERO) {
|
|
|
|
/*
|
|
|
|
* caches using SLAB_DESTROY_BY_RCU should let
|
|
|
|
* sk_node.next un-modified. Special care is taken
|
|
|
|
* when initializing object to zero.
|
|
|
|
*/
|
|
|
|
if (offsetof(struct sock, sk_node.next) != 0)
|
|
|
|
memset(sk, 0, offsetof(struct sock, sk_node.next));
|
|
|
|
memset(&sk->sk_node.pprev, 0,
|
|
|
|
prot->obj_size - offsetof(struct sock,
|
|
|
|
sk_node.pprev));
|
|
|
|
}
|
|
|
|
}
|
2007-11-01 08:33:50 +01:00
|
|
|
else
|
|
|
|
sk = kmalloc(prot->obj_size, priority);
|
|
|
|
|
2007-11-01 08:36:26 +01:00
|
|
|
if (sk != NULL) {
|
2009-02-26 14:46:57 +01:00
|
|
|
kmemcheck_annotate_bitfield(sk, flags);
|
|
|
|
|
2007-11-01 08:36:26 +01:00
|
|
|
if (security_sk_alloc(sk, family, priority))
|
|
|
|
goto out_free;
|
|
|
|
|
|
|
|
if (!try_module_get(prot->owner))
|
|
|
|
goto out_free_sec;
|
|
|
|
}
|
|
|
|
|
2007-11-01 08:33:50 +01:00
|
|
|
return sk;
|
2007-11-01 08:36:26 +01:00
|
|
|
|
|
|
|
out_free_sec:
|
|
|
|
security_sk_free(sk);
|
|
|
|
out_free:
|
|
|
|
if (slab != NULL)
|
|
|
|
kmem_cache_free(slab, sk);
|
|
|
|
else
|
|
|
|
kfree(sk);
|
|
|
|
return NULL;
|
2007-11-01 08:33:50 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
static void sk_prot_free(struct proto *prot, struct sock *sk)
|
|
|
|
{
|
|
|
|
struct kmem_cache *slab;
|
2007-11-01 08:36:26 +01:00
|
|
|
struct module *owner;
|
2007-11-01 08:33:50 +01:00
|
|
|
|
2007-11-01 08:36:26 +01:00
|
|
|
owner = prot->owner;
|
2007-11-01 08:33:50 +01:00
|
|
|
slab = prot->slab;
|
2007-11-01 08:36:26 +01:00
|
|
|
|
|
|
|
security_sk_free(sk);
|
2007-11-01 08:33:50 +01:00
|
|
|
if (slab != NULL)
|
|
|
|
kmem_cache_free(slab, sk);
|
|
|
|
else
|
|
|
|
kfree(sk);
|
2007-11-01 08:36:26 +01:00
|
|
|
module_put(owner);
|
2007-11-01 08:33:50 +01:00
|
|
|
}
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/**
|
|
|
|
* sk_alloc - All socket objects are allocated here
|
2007-10-13 06:17:49 +02:00
|
|
|
* @net: the applicable net namespace
|
2005-05-01 17:59:25 +02:00
|
|
|
* @family: protocol family
|
|
|
|
* @priority: for allocation (%GFP_KERNEL, %GFP_ATOMIC, etc)
|
|
|
|
* @prot: struct proto associated with this new sock instance
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
2007-10-09 08:24:22 +02:00
|
|
|
struct sock *sk_alloc(struct net *net, int family, gfp_t priority,
|
2007-11-01 08:39:31 +01:00
|
|
|
struct proto *prot)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2007-11-01 08:33:50 +01:00
|
|
|
struct sock *sk;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-11-01 08:38:43 +01:00
|
|
|
sk = sk_prot_alloc(prot, priority | __GFP_ZERO, family);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (sk) {
|
2007-11-01 08:38:43 +01:00
|
|
|
sk->sk_family = family;
|
|
|
|
/*
|
|
|
|
* See comment in struct sock definition to understand
|
|
|
|
* why we need sk_prot_creator -acme
|
|
|
|
*/
|
|
|
|
sk->sk_prot = sk->sk_prot_creator = prot;
|
|
|
|
sock_lock_init(sk);
|
2008-03-25 18:26:21 +01:00
|
|
|
sock_net_set(sk, get_net(net));
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2005-09-28 00:23:38 +02:00
|
|
|
|
2007-11-01 08:36:26 +01:00
|
|
|
return sk;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sk_alloc);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
net: No more expensive sock_hold()/sock_put() on each tx
One of the problem with sock memory accounting is it uses
a pair of sock_hold()/sock_put() for each transmitted packet.
This slows down bidirectional flows because the receive path
also needs to take a refcount on socket and might use a different
cpu than transmit path or transmit completion path. So these
two atomic operations also trigger cache line bounces.
We can see this in tx or tx/rx workloads (media gateways for example),
where sock_wfree() can be in top five functions in profiles.
We use this sock_hold()/sock_put() so that sock freeing
is delayed until all tx packets are completed.
As we also update sk_wmem_alloc, we could offset sk_wmem_alloc
by one unit at init time, until sk_free() is called.
Once sk_free() is called, we atomic_dec_and_test(sk_wmem_alloc)
to decrement initial offset and atomicaly check if any packets
are in flight.
skb_set_owner_w() doesnt call sock_hold() anymore
sock_wfree() doesnt call sock_put() anymore, but check if sk_wmem_alloc
reached 0 to perform the final freeing.
Drawback is that a skb->truesize error could lead to unfreeable sockets, or
even worse, prematurely calling __sk_free() on a live socket.
Nice speedups on SMP. tbench for example, going from 2691 MB/s to 2711 MB/s
on my 8 cpu dev machine, even if tbench was not really hitting sk_refcnt
contention point. 5 % speedup on a UDP transmit workload (depends
on number of flows), lowering TX completion cpu usage.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-11 11:55:43 +02:00
|
|
|
static void __sk_free(struct sock *sk)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct sk_filter *filter;
|
|
|
|
|
|
|
|
if (sk->sk_destruct)
|
|
|
|
sk->sk_destruct(sk);
|
|
|
|
|
2006-09-01 00:28:39 +02:00
|
|
|
filter = rcu_dereference(sk->sk_filter);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (filter) {
|
2007-10-18 06:21:51 +02:00
|
|
|
sk_filter_uncharge(sk, filter);
|
2006-09-01 00:28:39 +02:00
|
|
|
rcu_assign_pointer(sk->sk_filter, NULL);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2009-02-12 06:03:38 +01:00
|
|
|
sock_disable_timestamp(sk, SOCK_TIMESTAMP);
|
|
|
|
sock_disable_timestamp(sk, SOCK_TIMESTAMPING_RX_SOFTWARE);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (atomic_read(&sk->sk_omem_alloc))
|
|
|
|
printk(KERN_DEBUG "%s: optmem leakage (%d bytes) detected.\n",
|
2008-03-06 05:47:47 +01:00
|
|
|
__func__, atomic_read(&sk->sk_omem_alloc));
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2008-03-25 18:26:21 +01:00
|
|
|
put_net(sock_net(sk));
|
2007-11-01 08:33:50 +01:00
|
|
|
sk_prot_free(sk->sk_prot_creator, sk);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
net: No more expensive sock_hold()/sock_put() on each tx
One of the problem with sock memory accounting is it uses
a pair of sock_hold()/sock_put() for each transmitted packet.
This slows down bidirectional flows because the receive path
also needs to take a refcount on socket and might use a different
cpu than transmit path or transmit completion path. So these
two atomic operations also trigger cache line bounces.
We can see this in tx or tx/rx workloads (media gateways for example),
where sock_wfree() can be in top five functions in profiles.
We use this sock_hold()/sock_put() so that sock freeing
is delayed until all tx packets are completed.
As we also update sk_wmem_alloc, we could offset sk_wmem_alloc
by one unit at init time, until sk_free() is called.
Once sk_free() is called, we atomic_dec_and_test(sk_wmem_alloc)
to decrement initial offset and atomicaly check if any packets
are in flight.
skb_set_owner_w() doesnt call sock_hold() anymore
sock_wfree() doesnt call sock_put() anymore, but check if sk_wmem_alloc
reached 0 to perform the final freeing.
Drawback is that a skb->truesize error could lead to unfreeable sockets, or
even worse, prematurely calling __sk_free() on a live socket.
Nice speedups on SMP. tbench for example, going from 2691 MB/s to 2711 MB/s
on my 8 cpu dev machine, even if tbench was not really hitting sk_refcnt
contention point. 5 % speedup on a UDP transmit workload (depends
on number of flows), lowering TX completion cpu usage.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-11 11:55:43 +02:00
|
|
|
|
|
|
|
void sk_free(struct sock *sk)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* We substract one from sk_wmem_alloc and can know if
|
|
|
|
* some packets are still in some tx queue.
|
|
|
|
* If not null, sock_wfree() will call __sk_free(sk) later
|
|
|
|
*/
|
|
|
|
if (atomic_dec_and_test(&sk->sk_wmem_alloc))
|
|
|
|
__sk_free(sk);
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sk_free);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2008-02-29 20:18:32 +01:00
|
|
|
/*
|
|
|
|
* Last sock_put should drop referrence to sk->sk_net. It has already
|
|
|
|
* been dropped in sk_change_net. Taking referrence to stopping namespace
|
|
|
|
* is not an option.
|
|
|
|
* Take referrence to a socket to remove it from hash _alive_ and after that
|
|
|
|
* destroy it in the context of init_net.
|
|
|
|
*/
|
|
|
|
void sk_release_kernel(struct sock *sk)
|
|
|
|
{
|
|
|
|
if (sk == NULL || sk->sk_socket == NULL)
|
|
|
|
return;
|
|
|
|
|
|
|
|
sock_hold(sk);
|
|
|
|
sock_release(sk->sk_socket);
|
2008-04-16 10:59:46 +02:00
|
|
|
release_net(sock_net(sk));
|
2008-03-25 18:26:21 +01:00
|
|
|
sock_net_set(sk, get_net(&init_net));
|
2008-02-29 20:18:32 +01:00
|
|
|
sock_put(sk);
|
|
|
|
}
|
2008-02-29 20:33:19 +01:00
|
|
|
EXPORT_SYMBOL(sk_release_kernel);
|
2008-02-29 20:18:32 +01:00
|
|
|
|
2005-10-07 08:46:04 +02:00
|
|
|
struct sock *sk_clone(const struct sock *sk, const gfp_t priority)
|
2005-08-10 05:10:12 +02:00
|
|
|
{
|
2007-11-01 08:37:32 +01:00
|
|
|
struct sock *newsk;
|
2005-08-10 05:10:12 +02:00
|
|
|
|
2007-11-01 08:37:32 +01:00
|
|
|
newsk = sk_prot_alloc(sk->sk_prot, priority, sk->sk_family);
|
2005-08-10 05:10:12 +02:00
|
|
|
if (newsk != NULL) {
|
|
|
|
struct sk_filter *filter;
|
|
|
|
|
2006-08-05 08:08:56 +02:00
|
|
|
sock_copy(newsk, sk);
|
2005-08-10 05:10:12 +02:00
|
|
|
|
|
|
|
/* SANITY */
|
2008-03-25 18:26:21 +01:00
|
|
|
get_net(sock_net(newsk));
|
2005-08-10 05:10:12 +02:00
|
|
|
sk_node_init(&newsk->sk_node);
|
|
|
|
sock_lock_init(newsk);
|
|
|
|
bh_lock_sock(newsk);
|
2007-03-05 01:05:44 +01:00
|
|
|
newsk->sk_backlog.head = newsk->sk_backlog.tail = NULL;
|
2005-08-10 05:10:12 +02:00
|
|
|
|
|
|
|
atomic_set(&newsk->sk_rmem_alloc, 0);
|
net: No more expensive sock_hold()/sock_put() on each tx
One of the problem with sock memory accounting is it uses
a pair of sock_hold()/sock_put() for each transmitted packet.
This slows down bidirectional flows because the receive path
also needs to take a refcount on socket and might use a different
cpu than transmit path or transmit completion path. So these
two atomic operations also trigger cache line bounces.
We can see this in tx or tx/rx workloads (media gateways for example),
where sock_wfree() can be in top five functions in profiles.
We use this sock_hold()/sock_put() so that sock freeing
is delayed until all tx packets are completed.
As we also update sk_wmem_alloc, we could offset sk_wmem_alloc
by one unit at init time, until sk_free() is called.
Once sk_free() is called, we atomic_dec_and_test(sk_wmem_alloc)
to decrement initial offset and atomicaly check if any packets
are in flight.
skb_set_owner_w() doesnt call sock_hold() anymore
sock_wfree() doesnt call sock_put() anymore, but check if sk_wmem_alloc
reached 0 to perform the final freeing.
Drawback is that a skb->truesize error could lead to unfreeable sockets, or
even worse, prematurely calling __sk_free() on a live socket.
Nice speedups on SMP. tbench for example, going from 2691 MB/s to 2711 MB/s
on my 8 cpu dev machine, even if tbench was not really hitting sk_refcnt
contention point. 5 % speedup on a UDP transmit workload (depends
on number of flows), lowering TX completion cpu usage.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-11 11:55:43 +02:00
|
|
|
/*
|
|
|
|
* sk_wmem_alloc set to one (see sk_free() and sock_wfree())
|
|
|
|
*/
|
|
|
|
atomic_set(&newsk->sk_wmem_alloc, 1);
|
2005-08-10 05:10:12 +02:00
|
|
|
atomic_set(&newsk->sk_omem_alloc, 0);
|
|
|
|
skb_queue_head_init(&newsk->sk_receive_queue);
|
|
|
|
skb_queue_head_init(&newsk->sk_write_queue);
|
2006-05-24 02:55:33 +02:00
|
|
|
#ifdef CONFIG_NET_DMA
|
|
|
|
skb_queue_head_init(&newsk->sk_async_wait_queue);
|
|
|
|
#endif
|
2005-08-10 05:10:12 +02:00
|
|
|
|
|
|
|
rwlock_init(&newsk->sk_dst_lock);
|
|
|
|
rwlock_init(&newsk->sk_callback_lock);
|
2007-07-19 10:49:00 +02:00
|
|
|
lockdep_set_class_and_name(&newsk->sk_callback_lock,
|
|
|
|
af_callback_keys + newsk->sk_family,
|
|
|
|
af_family_clock_key_strings[newsk->sk_family]);
|
2005-08-10 05:10:12 +02:00
|
|
|
|
|
|
|
newsk->sk_dst_cache = NULL;
|
|
|
|
newsk->sk_wmem_queued = 0;
|
|
|
|
newsk->sk_forward_alloc = 0;
|
|
|
|
newsk->sk_send_head = NULL;
|
|
|
|
newsk->sk_userlocks = sk->sk_userlocks & ~SOCK_BINDPORT_LOCK;
|
|
|
|
|
|
|
|
sock_reset_flag(newsk, SOCK_DONE);
|
|
|
|
skb_queue_head_init(&newsk->sk_error_queue);
|
|
|
|
|
|
|
|
filter = newsk->sk_filter;
|
|
|
|
if (filter != NULL)
|
|
|
|
sk_filter_charge(newsk, filter);
|
|
|
|
|
|
|
|
if (unlikely(xfrm_sk_clone_policy(newsk))) {
|
|
|
|
/* It is still raw copy of parent, so invalidate
|
|
|
|
* destructor and make plain sk_free() */
|
|
|
|
newsk->sk_destruct = NULL;
|
|
|
|
sk_free(newsk);
|
|
|
|
newsk = NULL;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
newsk->sk_err = 0;
|
|
|
|
newsk->sk_priority = 0;
|
2009-07-16 01:13:10 +02:00
|
|
|
/*
|
|
|
|
* Before updating sk_refcnt, we must commit prior changes to memory
|
|
|
|
* (Documentation/RCU/rculist_nulls.txt for details)
|
|
|
|
*/
|
|
|
|
smp_wmb();
|
2005-08-10 05:10:12 +02:00
|
|
|
atomic_set(&newsk->sk_refcnt, 2);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Increment the counter in the same struct proto as the master
|
|
|
|
* sock (sk_refcnt_debug_inc uses newsk->sk_prot->socks, that
|
|
|
|
* is the same as sk->sk_prot->socks, as this field was copied
|
|
|
|
* with memcpy).
|
|
|
|
*
|
|
|
|
* This _changes_ the previous behaviour, where
|
|
|
|
* tcp_create_openreq_child always was incrementing the
|
|
|
|
* equivalent to tcp_prot->socks (inet_sock_nr), so this have
|
|
|
|
* to be taken into account in all callers. -acme
|
|
|
|
*/
|
|
|
|
sk_refcnt_debug_inc(newsk);
|
2008-06-18 07:41:38 +02:00
|
|
|
sk_set_socket(newsk, NULL);
|
2005-08-10 05:10:12 +02:00
|
|
|
newsk->sk_sleep = NULL;
|
|
|
|
|
|
|
|
if (newsk->sk_prot->sockets_allocated)
|
2008-11-26 06:16:35 +01:00
|
|
|
percpu_counter_inc(newsk->sk_prot->sockets_allocated);
|
2005-08-10 05:10:12 +02:00
|
|
|
}
|
|
|
|
out:
|
|
|
|
return newsk;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(sk_clone);
|
|
|
|
|
2007-04-21 02:12:43 +02:00
|
|
|
void sk_setup_caps(struct sock *sk, struct dst_entry *dst)
|
|
|
|
{
|
|
|
|
__sk_dst_set(sk, dst);
|
|
|
|
sk->sk_route_caps = dst->dev->features;
|
|
|
|
if (sk->sk_route_caps & NETIF_F_GSO)
|
2007-06-01 07:15:50 +02:00
|
|
|
sk->sk_route_caps |= NETIF_F_GSO_SOFTWARE;
|
2007-04-21 02:12:43 +02:00
|
|
|
if (sk_can_gso(sk)) {
|
[NET]: Add per-connection option to set max TSO frame size
Update: My mailer ate one of Jarek's feedback mails... Fixed the
parameter in netif_set_gso_max_size() to be u32, not u16. Fixed the
whitespace issue due to a patch import botch. Changed the types from
u32 to unsigned int to be more consistent with other variables in the
area. Also brought the patch up to the latest net-2.6.26 tree.
Update: Made gso_max_size container 32 bits, not 16. Moved the
location of gso_max_size within netdev to be less hotpath. Made more
consistent names between the sock and netdev layers, and added a
define for the max GSO size.
Update: Respun for net-2.6.26 tree.
Update: changed max_gso_frame_size and sk_gso_max_size from signed to
unsigned - thanks Stephen!
This patch adds the ability for device drivers to control the size of
the TSO frames being sent to them, per TCP connection. By setting the
netdevice's gso_max_size value, the socket layer will set the GSO
frame size based on that value. This will propogate into the TCP
layer, and send TSO's of that size to the hardware.
This can be desirable to help tune the bursty nature of TSO on a
per-adapter basis, where one may have 1 GbE and 10 GbE devices
coexisting in a system, one running multiqueue and the other not, etc.
This can also be desirable for devices that cannot support full 64 KB
TSO's, but still want to benefit from some level of segmentation
offloading.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2008-03-21 11:43:19 +01:00
|
|
|
if (dst->header_len) {
|
2007-04-21 02:12:43 +02:00
|
|
|
sk->sk_route_caps &= ~NETIF_F_GSO_MASK;
|
[NET]: Add per-connection option to set max TSO frame size
Update: My mailer ate one of Jarek's feedback mails... Fixed the
parameter in netif_set_gso_max_size() to be u32, not u16. Fixed the
whitespace issue due to a patch import botch. Changed the types from
u32 to unsigned int to be more consistent with other variables in the
area. Also brought the patch up to the latest net-2.6.26 tree.
Update: Made gso_max_size container 32 bits, not 16. Moved the
location of gso_max_size within netdev to be less hotpath. Made more
consistent names between the sock and netdev layers, and added a
define for the max GSO size.
Update: Respun for net-2.6.26 tree.
Update: changed max_gso_frame_size and sk_gso_max_size from signed to
unsigned - thanks Stephen!
This patch adds the ability for device drivers to control the size of
the TSO frames being sent to them, per TCP connection. By setting the
netdevice's gso_max_size value, the socket layer will set the GSO
frame size based on that value. This will propogate into the TCP
layer, and send TSO's of that size to the hardware.
This can be desirable to help tune the bursty nature of TSO on a
per-adapter basis, where one may have 1 GbE and 10 GbE devices
coexisting in a system, one running multiqueue and the other not, etc.
This can also be desirable for devices that cannot support full 64 KB
TSO's, but still want to benefit from some level of segmentation
offloading.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2008-03-21 11:43:19 +01:00
|
|
|
} else {
|
2007-04-21 02:12:43 +02:00
|
|
|
sk->sk_route_caps |= NETIF_F_SG | NETIF_F_HW_CSUM;
|
[NET]: Add per-connection option to set max TSO frame size
Update: My mailer ate one of Jarek's feedback mails... Fixed the
parameter in netif_set_gso_max_size() to be u32, not u16. Fixed the
whitespace issue due to a patch import botch. Changed the types from
u32 to unsigned int to be more consistent with other variables in the
area. Also brought the patch up to the latest net-2.6.26 tree.
Update: Made gso_max_size container 32 bits, not 16. Moved the
location of gso_max_size within netdev to be less hotpath. Made more
consistent names between the sock and netdev layers, and added a
define for the max GSO size.
Update: Respun for net-2.6.26 tree.
Update: changed max_gso_frame_size and sk_gso_max_size from signed to
unsigned - thanks Stephen!
This patch adds the ability for device drivers to control the size of
the TSO frames being sent to them, per TCP connection. By setting the
netdevice's gso_max_size value, the socket layer will set the GSO
frame size based on that value. This will propogate into the TCP
layer, and send TSO's of that size to the hardware.
This can be desirable to help tune the bursty nature of TSO on a
per-adapter basis, where one may have 1 GbE and 10 GbE devices
coexisting in a system, one running multiqueue and the other not, etc.
This can also be desirable for devices that cannot support full 64 KB
TSO's, but still want to benefit from some level of segmentation
offloading.
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2008-03-21 11:43:19 +01:00
|
|
|
sk->sk_gso_max_size = dst->dev->gso_max_size;
|
|
|
|
}
|
2007-04-21 02:12:43 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(sk_setup_caps);
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
void __init sk_init(void)
|
|
|
|
{
|
|
|
|
if (num_physpages <= 4096) {
|
|
|
|
sysctl_wmem_max = 32767;
|
|
|
|
sysctl_rmem_max = 32767;
|
|
|
|
sysctl_wmem_default = 32767;
|
|
|
|
sysctl_rmem_default = 32767;
|
|
|
|
} else if (num_physpages >= 131072) {
|
|
|
|
sysctl_wmem_max = 131071;
|
|
|
|
sysctl_rmem_max = 131071;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Simple resource managers for sockets.
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
2007-02-09 15:24:36 +01:00
|
|
|
/*
|
|
|
|
* Write buffer destructor automatically called from kfree_skb.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
|
|
|
void sock_wfree(struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
struct sock *sk = skb->sk;
|
net: No more expensive sock_hold()/sock_put() on each tx
One of the problem with sock memory accounting is it uses
a pair of sock_hold()/sock_put() for each transmitted packet.
This slows down bidirectional flows because the receive path
also needs to take a refcount on socket and might use a different
cpu than transmit path or transmit completion path. So these
two atomic operations also trigger cache line bounces.
We can see this in tx or tx/rx workloads (media gateways for example),
where sock_wfree() can be in top five functions in profiles.
We use this sock_hold()/sock_put() so that sock freeing
is delayed until all tx packets are completed.
As we also update sk_wmem_alloc, we could offset sk_wmem_alloc
by one unit at init time, until sk_free() is called.
Once sk_free() is called, we atomic_dec_and_test(sk_wmem_alloc)
to decrement initial offset and atomicaly check if any packets
are in flight.
skb_set_owner_w() doesnt call sock_hold() anymore
sock_wfree() doesnt call sock_put() anymore, but check if sk_wmem_alloc
reached 0 to perform the final freeing.
Drawback is that a skb->truesize error could lead to unfreeable sockets, or
even worse, prematurely calling __sk_free() on a live socket.
Nice speedups on SMP. tbench for example, going from 2691 MB/s to 2711 MB/s
on my 8 cpu dev machine, even if tbench was not really hitting sk_refcnt
contention point. 5 % speedup on a UDP transmit workload (depends
on number of flows), lowering TX completion cpu usage.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-11 11:55:43 +02:00
|
|
|
int res;
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/* In case it might be waiting for more memory. */
|
net: No more expensive sock_hold()/sock_put() on each tx
One of the problem with sock memory accounting is it uses
a pair of sock_hold()/sock_put() for each transmitted packet.
This slows down bidirectional flows because the receive path
also needs to take a refcount on socket and might use a different
cpu than transmit path or transmit completion path. So these
two atomic operations also trigger cache line bounces.
We can see this in tx or tx/rx workloads (media gateways for example),
where sock_wfree() can be in top five functions in profiles.
We use this sock_hold()/sock_put() so that sock freeing
is delayed until all tx packets are completed.
As we also update sk_wmem_alloc, we could offset sk_wmem_alloc
by one unit at init time, until sk_free() is called.
Once sk_free() is called, we atomic_dec_and_test(sk_wmem_alloc)
to decrement initial offset and atomicaly check if any packets
are in flight.
skb_set_owner_w() doesnt call sock_hold() anymore
sock_wfree() doesnt call sock_put() anymore, but check if sk_wmem_alloc
reached 0 to perform the final freeing.
Drawback is that a skb->truesize error could lead to unfreeable sockets, or
even worse, prematurely calling __sk_free() on a live socket.
Nice speedups on SMP. tbench for example, going from 2691 MB/s to 2711 MB/s
on my 8 cpu dev machine, even if tbench was not really hitting sk_refcnt
contention point. 5 % speedup on a UDP transmit workload (depends
on number of flows), lowering TX completion cpu usage.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-11 11:55:43 +02:00
|
|
|
res = atomic_sub_return(skb->truesize, &sk->sk_wmem_alloc);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!sock_flag(sk, SOCK_USE_WRITE_QUEUE))
|
|
|
|
sk->sk_write_space(sk);
|
net: No more expensive sock_hold()/sock_put() on each tx
One of the problem with sock memory accounting is it uses
a pair of sock_hold()/sock_put() for each transmitted packet.
This slows down bidirectional flows because the receive path
also needs to take a refcount on socket and might use a different
cpu than transmit path or transmit completion path. So these
two atomic operations also trigger cache line bounces.
We can see this in tx or tx/rx workloads (media gateways for example),
where sock_wfree() can be in top five functions in profiles.
We use this sock_hold()/sock_put() so that sock freeing
is delayed until all tx packets are completed.
As we also update sk_wmem_alloc, we could offset sk_wmem_alloc
by one unit at init time, until sk_free() is called.
Once sk_free() is called, we atomic_dec_and_test(sk_wmem_alloc)
to decrement initial offset and atomicaly check if any packets
are in flight.
skb_set_owner_w() doesnt call sock_hold() anymore
sock_wfree() doesnt call sock_put() anymore, but check if sk_wmem_alloc
reached 0 to perform the final freeing.
Drawback is that a skb->truesize error could lead to unfreeable sockets, or
even worse, prematurely calling __sk_free() on a live socket.
Nice speedups on SMP. tbench for example, going from 2691 MB/s to 2711 MB/s
on my 8 cpu dev machine, even if tbench was not really hitting sk_refcnt
contention point. 5 % speedup on a UDP transmit workload (depends
on number of flows), lowering TX completion cpu usage.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-11 11:55:43 +02:00
|
|
|
/*
|
|
|
|
* if sk_wmem_alloc reached 0, we are last user and should
|
|
|
|
* free this sock, as sk_free() call could not do it.
|
|
|
|
*/
|
|
|
|
if (res == 0)
|
|
|
|
__sk_free(sk);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_wfree);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-02-09 15:24:36 +01:00
|
|
|
/*
|
|
|
|
* Read buffer destructor automatically called from kfree_skb.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
|
|
|
void sock_rfree(struct sk_buff *skb)
|
|
|
|
{
|
|
|
|
struct sock *sk = skb->sk;
|
|
|
|
|
|
|
|
atomic_sub(skb->truesize, &sk->sk_rmem_alloc);
|
2007-12-31 09:11:19 +01:00
|
|
|
sk_mem_uncharge(skb->sk, skb->truesize);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_rfree);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
|
|
|
|
int sock_i_uid(struct sock *sk)
|
|
|
|
{
|
|
|
|
int uid;
|
|
|
|
|
|
|
|
read_lock(&sk->sk_callback_lock);
|
|
|
|
uid = sk->sk_socket ? SOCK_INODE(sk->sk_socket)->i_uid : 0;
|
|
|
|
read_unlock(&sk->sk_callback_lock);
|
|
|
|
return uid;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_i_uid);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
unsigned long sock_i_ino(struct sock *sk)
|
|
|
|
{
|
|
|
|
unsigned long ino;
|
|
|
|
|
|
|
|
read_lock(&sk->sk_callback_lock);
|
|
|
|
ino = sk->sk_socket ? SOCK_INODE(sk->sk_socket)->i_ino : 0;
|
|
|
|
read_unlock(&sk->sk_callback_lock);
|
|
|
|
return ino;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_i_ino);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Allocate a skb from the socket's send buffer.
|
|
|
|
*/
|
2005-07-08 23:57:47 +02:00
|
|
|
struct sk_buff *sock_wmalloc(struct sock *sk, unsigned long size, int force,
|
2005-10-07 08:46:04 +02:00
|
|
|
gfp_t priority)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if (force || atomic_read(&sk->sk_wmem_alloc) < sk->sk_sndbuf) {
|
2009-05-27 13:30:05 +02:00
|
|
|
struct sk_buff *skb = alloc_skb(size, priority);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (skb) {
|
|
|
|
skb_set_owner_w(skb, sk);
|
|
|
|
return skb;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_wmalloc);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Allocate a skb from the socket's receive buffer.
|
2007-02-09 15:24:36 +01:00
|
|
|
*/
|
2005-07-08 23:57:47 +02:00
|
|
|
struct sk_buff *sock_rmalloc(struct sock *sk, unsigned long size, int force,
|
2005-10-07 08:46:04 +02:00
|
|
|
gfp_t priority)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if (force || atomic_read(&sk->sk_rmem_alloc) < sk->sk_rcvbuf) {
|
|
|
|
struct sk_buff *skb = alloc_skb(size, priority);
|
|
|
|
if (skb) {
|
|
|
|
skb_set_owner_r(skb, sk);
|
|
|
|
return skb;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2007-02-09 15:24:36 +01:00
|
|
|
/*
|
2005-04-17 00:20:36 +02:00
|
|
|
* Allocate a memory block from the socket's option memory buffer.
|
2007-02-09 15:24:36 +01:00
|
|
|
*/
|
2005-10-07 08:46:04 +02:00
|
|
|
void *sock_kmalloc(struct sock *sk, int size, gfp_t priority)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
if ((unsigned)size <= sysctl_optmem_max &&
|
|
|
|
atomic_read(&sk->sk_omem_alloc) + size < sysctl_optmem_max) {
|
|
|
|
void *mem;
|
|
|
|
/* First do the add, to avoid the race if kmalloc
|
2007-02-09 15:24:36 +01:00
|
|
|
* might sleep.
|
2005-04-17 00:20:36 +02:00
|
|
|
*/
|
|
|
|
atomic_add(size, &sk->sk_omem_alloc);
|
|
|
|
mem = kmalloc(size, priority);
|
|
|
|
if (mem)
|
|
|
|
return mem;
|
|
|
|
atomic_sub(size, &sk->sk_omem_alloc);
|
|
|
|
}
|
|
|
|
return NULL;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_kmalloc);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Free an option memory block.
|
|
|
|
*/
|
|
|
|
void sock_kfree_s(struct sock *sk, void *mem, int size)
|
|
|
|
{
|
|
|
|
kfree(mem);
|
|
|
|
atomic_sub(size, &sk->sk_omem_alloc);
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_kfree_s);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/* It is almost wait_for_tcp_memory minus release_sock/lock_sock.
|
|
|
|
I think, these locks should be removed for datagram sockets.
|
|
|
|
*/
|
2009-05-27 13:30:05 +02:00
|
|
|
static long sock_wait_for_wmem(struct sock *sk, long timeo)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
DEFINE_WAIT(wait);
|
|
|
|
|
|
|
|
clear_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags);
|
|
|
|
for (;;) {
|
|
|
|
if (!timeo)
|
|
|
|
break;
|
|
|
|
if (signal_pending(current))
|
|
|
|
break;
|
|
|
|
set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
|
|
|
|
prepare_to_wait(sk->sk_sleep, &wait, TASK_INTERRUPTIBLE);
|
|
|
|
if (atomic_read(&sk->sk_wmem_alloc) < sk->sk_sndbuf)
|
|
|
|
break;
|
|
|
|
if (sk->sk_shutdown & SEND_SHUTDOWN)
|
|
|
|
break;
|
|
|
|
if (sk->sk_err)
|
|
|
|
break;
|
|
|
|
timeo = schedule_timeout(timeo);
|
|
|
|
}
|
|
|
|
finish_wait(sk->sk_sleep, &wait);
|
|
|
|
return timeo;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Generic send/receive buffer handlers
|
|
|
|
*/
|
|
|
|
|
2009-02-05 01:55:54 +01:00
|
|
|
struct sk_buff *sock_alloc_send_pskb(struct sock *sk, unsigned long header_len,
|
|
|
|
unsigned long data_len, int noblock,
|
|
|
|
int *errcode)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
struct sk_buff *skb;
|
2005-10-21 09:20:43 +02:00
|
|
|
gfp_t gfp_mask;
|
2005-04-17 00:20:36 +02:00
|
|
|
long timeo;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
gfp_mask = sk->sk_allocation;
|
|
|
|
if (gfp_mask & __GFP_WAIT)
|
|
|
|
gfp_mask |= __GFP_REPEAT;
|
|
|
|
|
|
|
|
timeo = sock_sndtimeo(sk, noblock);
|
|
|
|
while (1) {
|
|
|
|
err = sock_error(sk);
|
|
|
|
if (err != 0)
|
|
|
|
goto failure;
|
|
|
|
|
|
|
|
err = -EPIPE;
|
|
|
|
if (sk->sk_shutdown & SEND_SHUTDOWN)
|
|
|
|
goto failure;
|
|
|
|
|
|
|
|
if (atomic_read(&sk->sk_wmem_alloc) < sk->sk_sndbuf) {
|
2006-11-04 01:05:45 +01:00
|
|
|
skb = alloc_skb(header_len, gfp_mask);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (skb) {
|
|
|
|
int npages;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/* No pages, we're done... */
|
|
|
|
if (!data_len)
|
|
|
|
break;
|
|
|
|
|
|
|
|
npages = (data_len + (PAGE_SIZE - 1)) >> PAGE_SHIFT;
|
|
|
|
skb->truesize += data_len;
|
|
|
|
skb_shinfo(skb)->nr_frags = npages;
|
|
|
|
for (i = 0; i < npages; i++) {
|
|
|
|
struct page *page;
|
|
|
|
skb_frag_t *frag;
|
|
|
|
|
|
|
|
page = alloc_pages(sk->sk_allocation, 0);
|
|
|
|
if (!page) {
|
|
|
|
err = -ENOBUFS;
|
|
|
|
skb_shinfo(skb)->nr_frags = i;
|
|
|
|
kfree_skb(skb);
|
|
|
|
goto failure;
|
|
|
|
}
|
|
|
|
|
|
|
|
frag = &skb_shinfo(skb)->frags[i];
|
|
|
|
frag->page = page;
|
|
|
|
frag->page_offset = 0;
|
|
|
|
frag->size = (data_len >= PAGE_SIZE ?
|
|
|
|
PAGE_SIZE :
|
|
|
|
data_len);
|
|
|
|
data_len -= PAGE_SIZE;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Full success... */
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
err = -ENOBUFS;
|
|
|
|
goto failure;
|
|
|
|
}
|
|
|
|
set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags);
|
|
|
|
set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
|
|
|
|
err = -EAGAIN;
|
|
|
|
if (!timeo)
|
|
|
|
goto failure;
|
|
|
|
if (signal_pending(current))
|
|
|
|
goto interrupted;
|
|
|
|
timeo = sock_wait_for_wmem(sk, timeo);
|
|
|
|
}
|
|
|
|
|
|
|
|
skb_set_owner_w(skb, sk);
|
|
|
|
return skb;
|
|
|
|
|
|
|
|
interrupted:
|
|
|
|
err = sock_intr_errno(timeo);
|
|
|
|
failure:
|
|
|
|
*errcode = err;
|
|
|
|
return NULL;
|
|
|
|
}
|
2009-02-05 01:55:54 +01:00
|
|
|
EXPORT_SYMBOL(sock_alloc_send_pskb);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-02-09 15:24:36 +01:00
|
|
|
struct sk_buff *sock_alloc_send_skb(struct sock *sk, unsigned long size,
|
2005-04-17 00:20:36 +02:00
|
|
|
int noblock, int *errcode)
|
|
|
|
{
|
|
|
|
return sock_alloc_send_pskb(sk, size, 0, noblock, errcode);
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_alloc_send_skb);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
static void __lock_sock(struct sock *sk)
|
|
|
|
{
|
|
|
|
DEFINE_WAIT(wait);
|
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
for (;;) {
|
2005-04-17 00:20:36 +02:00
|
|
|
prepare_to_wait_exclusive(&sk->sk_lock.wq, &wait,
|
|
|
|
TASK_UNINTERRUPTIBLE);
|
|
|
|
spin_unlock_bh(&sk->sk_lock.slock);
|
|
|
|
schedule();
|
|
|
|
spin_lock_bh(&sk->sk_lock.slock);
|
2007-04-11 05:10:33 +02:00
|
|
|
if (!sock_owned_by_user(sk))
|
2005-04-17 00:20:36 +02:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
finish_wait(&sk->sk_lock.wq, &wait);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __release_sock(struct sock *sk)
|
|
|
|
{
|
|
|
|
struct sk_buff *skb = sk->sk_backlog.head;
|
|
|
|
|
|
|
|
do {
|
|
|
|
sk->sk_backlog.head = sk->sk_backlog.tail = NULL;
|
|
|
|
bh_unlock_sock(sk);
|
|
|
|
|
|
|
|
do {
|
|
|
|
struct sk_buff *next = skb->next;
|
|
|
|
|
|
|
|
skb->next = NULL;
|
2008-10-07 23:18:42 +02:00
|
|
|
sk_backlog_rcv(sk, skb);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* We are in process context here with softirqs
|
|
|
|
* disabled, use cond_resched_softirq() to preempt.
|
|
|
|
* This is safe to do because we've taken the backlog
|
|
|
|
* queue private:
|
|
|
|
*/
|
|
|
|
cond_resched_softirq();
|
|
|
|
|
|
|
|
skb = next;
|
|
|
|
} while (skb != NULL);
|
|
|
|
|
|
|
|
bh_lock_sock(sk);
|
2007-04-11 05:10:33 +02:00
|
|
|
} while ((skb = sk->sk_backlog.head) != NULL);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* sk_wait_data - wait for data to arrive at sk_receive_queue
|
2005-05-01 17:59:25 +02:00
|
|
|
* @sk: sock to wait on
|
|
|
|
* @timeo: for how long
|
2005-04-17 00:20:36 +02:00
|
|
|
*
|
|
|
|
* Now socket state including sk->sk_err is changed only under lock,
|
|
|
|
* hence we may omit checks after joining wait queue.
|
|
|
|
* We check receive queue before schedule() only as optimization;
|
|
|
|
* it is very likely that release_sock() added new data.
|
|
|
|
*/
|
|
|
|
int sk_wait_data(struct sock *sk, long *timeo)
|
|
|
|
{
|
|
|
|
int rc;
|
|
|
|
DEFINE_WAIT(wait);
|
|
|
|
|
|
|
|
prepare_to_wait(sk->sk_sleep, &wait, TASK_INTERRUPTIBLE);
|
|
|
|
set_bit(SOCK_ASYNC_WAITDATA, &sk->sk_socket->flags);
|
|
|
|
rc = sk_wait_event(sk, timeo, !skb_queue_empty(&sk->sk_receive_queue));
|
|
|
|
clear_bit(SOCK_ASYNC_WAITDATA, &sk->sk_socket->flags);
|
|
|
|
finish_wait(sk->sk_sleep, &wait);
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(sk_wait_data);
|
|
|
|
|
2007-12-31 09:11:19 +01:00
|
|
|
/**
|
|
|
|
* __sk_mem_schedule - increase sk_forward_alloc and memory_allocated
|
|
|
|
* @sk: socket
|
|
|
|
* @size: memory size to allocate
|
|
|
|
* @kind: allocation type
|
|
|
|
*
|
|
|
|
* If kind is SK_MEM_SEND, it means wmem allocation. Otherwise it means
|
|
|
|
* rmem allocation. This function assumes that protocols which have
|
|
|
|
* memory_pressure use sk_wmem_queued as write buffer accounting.
|
|
|
|
*/
|
|
|
|
int __sk_mem_schedule(struct sock *sk, int size, int kind)
|
|
|
|
{
|
|
|
|
struct proto *prot = sk->sk_prot;
|
|
|
|
int amt = sk_mem_pages(size);
|
|
|
|
int allocated;
|
|
|
|
|
|
|
|
sk->sk_forward_alloc += amt * SK_MEM_QUANTUM;
|
|
|
|
allocated = atomic_add_return(amt, prot->memory_allocated);
|
|
|
|
|
|
|
|
/* Under limit. */
|
|
|
|
if (allocated <= prot->sysctl_mem[0]) {
|
|
|
|
if (prot->memory_pressure && *prot->memory_pressure)
|
|
|
|
*prot->memory_pressure = 0;
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Under pressure. */
|
|
|
|
if (allocated > prot->sysctl_mem[1])
|
|
|
|
if (prot->enter_memory_pressure)
|
2008-07-17 05:28:10 +02:00
|
|
|
prot->enter_memory_pressure(sk);
|
2007-12-31 09:11:19 +01:00
|
|
|
|
|
|
|
/* Over hard limit. */
|
|
|
|
if (allocated > prot->sysctl_mem[2])
|
|
|
|
goto suppress_allocation;
|
|
|
|
|
|
|
|
/* guarantee minimum buffer size under pressure */
|
|
|
|
if (kind == SK_MEM_RECV) {
|
|
|
|
if (atomic_read(&sk->sk_rmem_alloc) < prot->sysctl_rmem[0])
|
|
|
|
return 1;
|
|
|
|
} else { /* SK_MEM_SEND */
|
|
|
|
if (sk->sk_type == SOCK_STREAM) {
|
|
|
|
if (sk->sk_wmem_queued < prot->sysctl_wmem[0])
|
|
|
|
return 1;
|
|
|
|
} else if (atomic_read(&sk->sk_wmem_alloc) <
|
|
|
|
prot->sysctl_wmem[0])
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (prot->memory_pressure) {
|
2008-11-26 06:16:35 +01:00
|
|
|
int alloc;
|
|
|
|
|
|
|
|
if (!*prot->memory_pressure)
|
|
|
|
return 1;
|
|
|
|
alloc = percpu_counter_read_positive(prot->sockets_allocated);
|
|
|
|
if (prot->sysctl_mem[2] > alloc *
|
2007-12-31 09:11:19 +01:00
|
|
|
sk_mem_pages(sk->sk_wmem_queued +
|
|
|
|
atomic_read(&sk->sk_rmem_alloc) +
|
|
|
|
sk->sk_forward_alloc))
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
suppress_allocation:
|
|
|
|
|
|
|
|
if (kind == SK_MEM_SEND && sk->sk_type == SOCK_STREAM) {
|
|
|
|
sk_stream_moderate_sndbuf(sk);
|
|
|
|
|
|
|
|
/* Fail only if socket is _under_ its sndbuf.
|
|
|
|
* In this case we cannot block, so that we have to fail.
|
|
|
|
*/
|
|
|
|
if (sk->sk_wmem_queued + size >= sk->sk_sndbuf)
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Alas. Undo changes. */
|
|
|
|
sk->sk_forward_alloc -= amt * SK_MEM_QUANTUM;
|
|
|
|
atomic_sub(amt, prot->memory_allocated);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(__sk_mem_schedule);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* __sk_reclaim - reclaim memory_allocated
|
|
|
|
* @sk: socket
|
|
|
|
*/
|
|
|
|
void __sk_mem_reclaim(struct sock *sk)
|
|
|
|
{
|
|
|
|
struct proto *prot = sk->sk_prot;
|
|
|
|
|
2008-01-01 00:00:50 +01:00
|
|
|
atomic_sub(sk->sk_forward_alloc >> SK_MEM_QUANTUM_SHIFT,
|
2007-12-31 09:11:19 +01:00
|
|
|
prot->memory_allocated);
|
|
|
|
sk->sk_forward_alloc &= SK_MEM_QUANTUM - 1;
|
|
|
|
|
|
|
|
if (prot->memory_pressure && *prot->memory_pressure &&
|
|
|
|
(atomic_read(prot->memory_allocated) < prot->sysctl_mem[0]))
|
|
|
|
*prot->memory_pressure = 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(__sk_mem_reclaim);
|
|
|
|
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
/*
|
|
|
|
* 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.
|
|
|
|
*/
|
|
|
|
|
|
|
|
int sock_no_bind(struct socket *sock, struct sockaddr *saddr, int len)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_bind);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-02-09 15:24:36 +01:00
|
|
|
int sock_no_connect(struct socket *sock, struct sockaddr *saddr,
|
2005-04-17 00:20:36 +02:00
|
|
|
int len, int flags)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_connect);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int sock_no_socketpair(struct socket *sock1, struct socket *sock2)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_socketpair);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int sock_no_accept(struct socket *sock, struct socket *newsock, int flags)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_accept);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2007-02-09 15:24:36 +01:00
|
|
|
int sock_no_getname(struct socket *sock, struct sockaddr *saddr,
|
2005-04-17 00:20:36 +02:00
|
|
|
int *len, int peer)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_getname);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2009-05-27 13:30:05 +02:00
|
|
|
unsigned int sock_no_poll(struct file *file, struct socket *sock, poll_table *pt)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_poll);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int sock_no_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_ioctl);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int sock_no_listen(struct socket *sock, int backlog)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_listen);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int sock_no_shutdown(struct socket *sock, int how)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_shutdown);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int sock_no_setsockopt(struct socket *sock, int level, int optname,
|
|
|
|
char __user *optval, int optlen)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_setsockopt);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int sock_no_getsockopt(struct socket *sock, int level, int optname,
|
|
|
|
char __user *optval, int __user *optlen)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_getsockopt);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int sock_no_sendmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *m,
|
|
|
|
size_t len)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_sendmsg);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int sock_no_recvmsg(struct kiocb *iocb, struct socket *sock, struct msghdr *m,
|
|
|
|
size_t len, int flags)
|
|
|
|
{
|
|
|
|
return -EOPNOTSUPP;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_recvmsg);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
int sock_no_mmap(struct file *file, struct socket *sock, struct vm_area_struct *vma)
|
|
|
|
{
|
|
|
|
/* Mirror missing mmap method error code */
|
|
|
|
return -ENODEV;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_mmap);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
ssize_t sock_no_sendpage(struct socket *sock, struct page *page, int offset, size_t size, int flags)
|
|
|
|
{
|
|
|
|
ssize_t res;
|
|
|
|
struct msghdr msg = {.msg_flags = flags};
|
|
|
|
struct kvec iov;
|
|
|
|
char *kaddr = kmap(page);
|
|
|
|
iov.iov_base = kaddr + offset;
|
|
|
|
iov.iov_len = size;
|
|
|
|
res = kernel_sendmsg(sock, &msg, &iov, 1, size);
|
|
|
|
kunmap(page);
|
|
|
|
return res;
|
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_no_sendpage);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Default Socket Callbacks
|
|
|
|
*/
|
|
|
|
|
|
|
|
static void sock_def_wakeup(struct sock *sk)
|
|
|
|
{
|
|
|
|
read_lock(&sk->sk_callback_lock);
|
2009-07-08 14:09:13 +02:00
|
|
|
if (sk_has_sleeper(sk))
|
2005-04-17 00:20:36 +02:00
|
|
|
wake_up_interruptible_all(sk->sk_sleep);
|
|
|
|
read_unlock(&sk->sk_callback_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void sock_def_error_report(struct sock *sk)
|
|
|
|
{
|
|
|
|
read_lock(&sk->sk_callback_lock);
|
2009-07-08 14:09:13 +02:00
|
|
|
if (sk_has_sleeper(sk))
|
2009-04-01 00:24:21 +02:00
|
|
|
wake_up_interruptible_poll(sk->sk_sleep, POLLERR);
|
2007-11-26 13:10:50 +01:00
|
|
|
sk_wake_async(sk, SOCK_WAKE_IO, POLL_ERR);
|
2005-04-17 00:20:36 +02:00
|
|
|
read_unlock(&sk->sk_callback_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void sock_def_readable(struct sock *sk, int len)
|
|
|
|
{
|
|
|
|
read_lock(&sk->sk_callback_lock);
|
2009-07-08 14:09:13 +02:00
|
|
|
if (sk_has_sleeper(sk))
|
2009-04-01 00:24:21 +02:00
|
|
|
wake_up_interruptible_sync_poll(sk->sk_sleep, POLLIN |
|
|
|
|
POLLRDNORM | POLLRDBAND);
|
2007-11-26 13:10:50 +01:00
|
|
|
sk_wake_async(sk, SOCK_WAKE_WAITD, POLL_IN);
|
2005-04-17 00:20:36 +02:00
|
|
|
read_unlock(&sk->sk_callback_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void sock_def_write_space(struct sock *sk)
|
|
|
|
{
|
|
|
|
read_lock(&sk->sk_callback_lock);
|
|
|
|
|
|
|
|
/* Do not wake up a writer until he can make "significant"
|
|
|
|
* progress. --DaveM
|
|
|
|
*/
|
2007-04-11 05:10:33 +02:00
|
|
|
if ((atomic_read(&sk->sk_wmem_alloc) << 1) <= sk->sk_sndbuf) {
|
2009-07-08 14:09:13 +02:00
|
|
|
if (sk_has_sleeper(sk))
|
2009-04-01 00:24:21 +02:00
|
|
|
wake_up_interruptible_sync_poll(sk->sk_sleep, POLLOUT |
|
|
|
|
POLLWRNORM | POLLWRBAND);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
/* Should agree with poll, otherwise some programs break */
|
|
|
|
if (sock_writeable(sk))
|
2007-11-26 13:10:50 +01:00
|
|
|
sk_wake_async(sk, SOCK_WAKE_SPACE, POLL_OUT);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
read_unlock(&sk->sk_callback_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void sock_def_destruct(struct sock *sk)
|
|
|
|
{
|
2005-11-08 18:41:34 +01:00
|
|
|
kfree(sk->sk_protinfo);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
void sk_send_sigurg(struct sock *sk)
|
|
|
|
{
|
|
|
|
if (sk->sk_socket && sk->sk_socket->file)
|
|
|
|
if (send_sigurg(&sk->sk_socket->file->f_owner))
|
2007-11-26 13:10:50 +01:00
|
|
|
sk_wake_async(sk, SOCK_WAKE_URG, POLL_PRI);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sk_send_sigurg);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
void sk_reset_timer(struct sock *sk, struct timer_list* timer,
|
|
|
|
unsigned long expires)
|
|
|
|
{
|
|
|
|
if (!mod_timer(timer, expires))
|
|
|
|
sock_hold(sk);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(sk_reset_timer);
|
|
|
|
|
|
|
|
void sk_stop_timer(struct sock *sk, struct timer_list* timer)
|
|
|
|
{
|
|
|
|
if (timer_pending(timer) && del_timer(timer))
|
|
|
|
__sock_put(sk);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(sk_stop_timer);
|
|
|
|
|
|
|
|
void sock_init_data(struct socket *sock, struct sock *sk)
|
|
|
|
{
|
|
|
|
skb_queue_head_init(&sk->sk_receive_queue);
|
|
|
|
skb_queue_head_init(&sk->sk_write_queue);
|
|
|
|
skb_queue_head_init(&sk->sk_error_queue);
|
2006-05-24 02:55:33 +02:00
|
|
|
#ifdef CONFIG_NET_DMA
|
|
|
|
skb_queue_head_init(&sk->sk_async_wait_queue);
|
|
|
|
#endif
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
sk->sk_send_head = NULL;
|
|
|
|
|
|
|
|
init_timer(&sk->sk_timer);
|
2007-02-09 15:24:36 +01:00
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
sk->sk_allocation = GFP_KERNEL;
|
|
|
|
sk->sk_rcvbuf = sysctl_rmem_default;
|
|
|
|
sk->sk_sndbuf = sysctl_wmem_default;
|
|
|
|
sk->sk_state = TCP_CLOSE;
|
2008-06-18 07:41:38 +02:00
|
|
|
sk_set_socket(sk, sock);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
sock_set_flag(sk, SOCK_ZAPPED);
|
|
|
|
|
2007-04-11 05:10:33 +02:00
|
|
|
if (sock) {
|
2005-04-17 00:20:36 +02:00
|
|
|
sk->sk_type = sock->type;
|
|
|
|
sk->sk_sleep = &sock->wait;
|
|
|
|
sock->sk = sk;
|
|
|
|
} else
|
|
|
|
sk->sk_sleep = NULL;
|
|
|
|
|
|
|
|
rwlock_init(&sk->sk_dst_lock);
|
|
|
|
rwlock_init(&sk->sk_callback_lock);
|
2007-07-19 10:49:00 +02:00
|
|
|
lockdep_set_class_and_name(&sk->sk_callback_lock,
|
|
|
|
af_callback_keys + sk->sk_family,
|
|
|
|
af_family_clock_key_strings[sk->sk_family]);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
sk->sk_state_change = sock_def_wakeup;
|
|
|
|
sk->sk_data_ready = sock_def_readable;
|
|
|
|
sk->sk_write_space = sock_def_write_space;
|
|
|
|
sk->sk_error_report = sock_def_error_report;
|
|
|
|
sk->sk_destruct = sock_def_destruct;
|
|
|
|
|
|
|
|
sk->sk_sndmsg_page = NULL;
|
|
|
|
sk->sk_sndmsg_off = 0;
|
|
|
|
|
|
|
|
sk->sk_peercred.pid = 0;
|
|
|
|
sk->sk_peercred.uid = -1;
|
|
|
|
sk->sk_peercred.gid = -1;
|
|
|
|
sk->sk_write_pending = 0;
|
|
|
|
sk->sk_rcvlowat = 1;
|
|
|
|
sk->sk_rcvtimeo = MAX_SCHEDULE_TIMEOUT;
|
|
|
|
sk->sk_sndtimeo = MAX_SCHEDULE_TIMEOUT;
|
|
|
|
|
2008-04-14 06:39:26 +02:00
|
|
|
sk->sk_stamp = ktime_set(-1L, 0);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2009-07-16 01:13:10 +02:00
|
|
|
/*
|
|
|
|
* Before updating sk_refcnt, we must commit prior changes to memory
|
|
|
|
* (Documentation/RCU/rculist_nulls.txt for details)
|
|
|
|
*/
|
|
|
|
smp_wmb();
|
2005-04-17 00:20:36 +02:00
|
|
|
atomic_set(&sk->sk_refcnt, 1);
|
net: No more expensive sock_hold()/sock_put() on each tx
One of the problem with sock memory accounting is it uses
a pair of sock_hold()/sock_put() for each transmitted packet.
This slows down bidirectional flows because the receive path
also needs to take a refcount on socket and might use a different
cpu than transmit path or transmit completion path. So these
two atomic operations also trigger cache line bounces.
We can see this in tx or tx/rx workloads (media gateways for example),
where sock_wfree() can be in top five functions in profiles.
We use this sock_hold()/sock_put() so that sock freeing
is delayed until all tx packets are completed.
As we also update sk_wmem_alloc, we could offset sk_wmem_alloc
by one unit at init time, until sk_free() is called.
Once sk_free() is called, we atomic_dec_and_test(sk_wmem_alloc)
to decrement initial offset and atomicaly check if any packets
are in flight.
skb_set_owner_w() doesnt call sock_hold() anymore
sock_wfree() doesnt call sock_put() anymore, but check if sk_wmem_alloc
reached 0 to perform the final freeing.
Drawback is that a skb->truesize error could lead to unfreeable sockets, or
even worse, prematurely calling __sk_free() on a live socket.
Nice speedups on SMP. tbench for example, going from 2691 MB/s to 2711 MB/s
on my 8 cpu dev machine, even if tbench was not really hitting sk_refcnt
contention point. 5 % speedup on a UDP transmit workload (depends
on number of flows), lowering TX completion cpu usage.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2009-06-11 11:55:43 +02:00
|
|
|
atomic_set(&sk->sk_wmem_alloc, 1);
|
2007-11-14 05:30:01 +01:00
|
|
|
atomic_set(&sk->sk_drops, 0);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2009-05-27 13:30:05 +02:00
|
|
|
EXPORT_SYMBOL(sock_init_data);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2008-02-14 00:03:16 +01:00
|
|
|
void lock_sock_nested(struct sock *sk, int subclass)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
might_sleep();
|
2006-07-03 09:25:35 +02:00
|
|
|
spin_lock_bh(&sk->sk_lock.slock);
|
2007-09-12 10:44:19 +02:00
|
|
|
if (sk->sk_lock.owned)
|
2005-04-17 00:20:36 +02:00
|
|
|
__lock_sock(sk);
|
2007-09-12 10:44:19 +02:00
|
|
|
sk->sk_lock.owned = 1;
|
2006-07-03 09:25:35 +02:00
|
|
|
spin_unlock(&sk->sk_lock.slock);
|
|
|
|
/*
|
|
|
|
* The sk_lock has mutex_lock() semantics here:
|
|
|
|
*/
|
2006-11-09 07:44:35 +01:00
|
|
|
mutex_acquire(&sk->sk_lock.dep_map, subclass, 0, _RET_IP_);
|
2006-07-03 09:25:35 +02:00
|
|
|
local_bh_enable();
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
2006-11-09 07:44:35 +01:00
|
|
|
EXPORT_SYMBOL(lock_sock_nested);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
2008-02-14 00:03:16 +01:00
|
|
|
void release_sock(struct sock *sk)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
2006-07-03 09:25:35 +02:00
|
|
|
/*
|
|
|
|
* The sk_lock has mutex_unlock() semantics:
|
|
|
|
*/
|
|
|
|
mutex_release(&sk->sk_lock.dep_map, 1, _RET_IP_);
|
|
|
|
|
|
|
|
spin_lock_bh(&sk->sk_lock.slock);
|
2005-04-17 00:20:36 +02:00
|
|
|
if (sk->sk_backlog.tail)
|
|
|
|
__release_sock(sk);
|
2007-09-12 10:44:19 +02:00
|
|
|
sk->sk_lock.owned = 0;
|
2006-07-03 09:25:35 +02:00
|
|
|
if (waitqueue_active(&sk->sk_lock.wq))
|
|
|
|
wake_up(&sk->sk_lock.wq);
|
|
|
|
spin_unlock_bh(&sk->sk_lock.slock);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(release_sock);
|
|
|
|
|
|
|
|
int sock_get_timestamp(struct sock *sk, struct timeval __user *userstamp)
|
2007-02-09 15:24:36 +01:00
|
|
|
{
|
2007-04-20 01:16:32 +02:00
|
|
|
struct timeval tv;
|
2005-04-17 00:20:36 +02:00
|
|
|
if (!sock_flag(sk, SOCK_TIMESTAMP))
|
2009-02-12 06:03:38 +01:00
|
|
|
sock_enable_timestamp(sk, SOCK_TIMESTAMP);
|
2007-04-20 01:16:32 +02:00
|
|
|
tv = ktime_to_timeval(sk->sk_stamp);
|
|
|
|
if (tv.tv_sec == -1)
|
2005-04-17 00:20:36 +02:00
|
|
|
return -ENOENT;
|
2007-04-20 01:16:32 +02:00
|
|
|
if (tv.tv_sec == 0) {
|
|
|
|
sk->sk_stamp = ktime_get_real();
|
|
|
|
tv = ktime_to_timeval(sk->sk_stamp);
|
|
|
|
}
|
|
|
|
return copy_to_user(userstamp, &tv, sizeof(tv)) ? -EFAULT : 0;
|
2007-02-09 15:24:36 +01:00
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
EXPORT_SYMBOL(sock_get_timestamp);
|
|
|
|
|
2007-03-19 01:33:16 +01:00
|
|
|
int sock_get_timestampns(struct sock *sk, struct timespec __user *userstamp)
|
|
|
|
{
|
|
|
|
struct timespec ts;
|
|
|
|
if (!sock_flag(sk, SOCK_TIMESTAMP))
|
2009-02-12 06:03:38 +01:00
|
|
|
sock_enable_timestamp(sk, SOCK_TIMESTAMP);
|
2007-03-19 01:33:16 +01:00
|
|
|
ts = ktime_to_timespec(sk->sk_stamp);
|
|
|
|
if (ts.tv_sec == -1)
|
|
|
|
return -ENOENT;
|
|
|
|
if (ts.tv_sec == 0) {
|
|
|
|
sk->sk_stamp = ktime_get_real();
|
|
|
|
ts = ktime_to_timespec(sk->sk_stamp);
|
|
|
|
}
|
|
|
|
return copy_to_user(userstamp, &ts, sizeof(ts)) ? -EFAULT : 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(sock_get_timestampns);
|
|
|
|
|
2009-02-12 06:03:38 +01:00
|
|
|
void sock_enable_timestamp(struct sock *sk, int flag)
|
2007-02-09 15:24:36 +01:00
|
|
|
{
|
2009-02-12 06:03:38 +01:00
|
|
|
if (!sock_flag(sk, flag)) {
|
|
|
|
sock_set_flag(sk, flag);
|
|
|
|
/*
|
|
|
|
* we just set one of the two flags which require net
|
|
|
|
* time stamping, but time stamping might have been on
|
|
|
|
* already because of the other one
|
|
|
|
*/
|
|
|
|
if (!sock_flag(sk,
|
|
|
|
flag == SOCK_TIMESTAMP ?
|
|
|
|
SOCK_TIMESTAMPING_RX_SOFTWARE :
|
|
|
|
SOCK_TIMESTAMP))
|
|
|
|
net_enable_timestamp();
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Get a socket option on an socket.
|
|
|
|
*
|
|
|
|
* FIX: POSIX 1003.1g is very ambiguous here. It states that
|
|
|
|
* asynchronous errors should be reported by getsockopt. We assume
|
|
|
|
* this means if you specify SO_ERROR (otherwise whats the point of it).
|
|
|
|
*/
|
|
|
|
int sock_common_getsockopt(struct socket *sock, int level, int optname,
|
|
|
|
char __user *optval, int __user *optlen)
|
|
|
|
{
|
|
|
|
struct sock *sk = sock->sk;
|
|
|
|
|
|
|
|
return sk->sk_prot->getsockopt(sk, level, optname, optval, optlen);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(sock_common_getsockopt);
|
|
|
|
|
2006-03-21 07:45:21 +01:00
|
|
|
#ifdef CONFIG_COMPAT
|
2006-03-21 07:48:35 +01:00
|
|
|
int compat_sock_common_getsockopt(struct socket *sock, int level, int optname,
|
|
|
|
char __user *optval, int __user *optlen)
|
2006-03-21 07:45:21 +01:00
|
|
|
{
|
|
|
|
struct sock *sk = sock->sk;
|
|
|
|
|
2007-03-06 22:44:06 +01:00
|
|
|
if (sk->sk_prot->compat_getsockopt != NULL)
|
2006-03-21 07:48:35 +01:00
|
|
|
return sk->sk_prot->compat_getsockopt(sk, level, optname,
|
|
|
|
optval, optlen);
|
2006-03-21 07:45:21 +01:00
|
|
|
return sk->sk_prot->getsockopt(sk, level, optname, optval, optlen);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(compat_sock_common_getsockopt);
|
|
|
|
#endif
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
int sock_common_recvmsg(struct kiocb *iocb, struct socket *sock,
|
|
|
|
struct msghdr *msg, size_t size, int flags)
|
|
|
|
{
|
|
|
|
struct sock *sk = sock->sk;
|
|
|
|
int addr_len = 0;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
err = sk->sk_prot->recvmsg(iocb, sk, msg, size, flags & MSG_DONTWAIT,
|
|
|
|
flags & ~MSG_DONTWAIT, &addr_len);
|
|
|
|
if (err >= 0)
|
|
|
|
msg->msg_namelen = addr_len;
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(sock_common_recvmsg);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Set socket options on an inet socket.
|
|
|
|
*/
|
|
|
|
int sock_common_setsockopt(struct socket *sock, int level, int optname,
|
|
|
|
char __user *optval, int optlen)
|
|
|
|
{
|
|
|
|
struct sock *sk = sock->sk;
|
|
|
|
|
|
|
|
return sk->sk_prot->setsockopt(sk, level, optname, optval, optlen);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(sock_common_setsockopt);
|
|
|
|
|
2006-03-21 07:45:21 +01:00
|
|
|
#ifdef CONFIG_COMPAT
|
2006-03-21 07:48:35 +01:00
|
|
|
int compat_sock_common_setsockopt(struct socket *sock, int level, int optname,
|
|
|
|
char __user *optval, int optlen)
|
2006-03-21 07:45:21 +01:00
|
|
|
{
|
|
|
|
struct sock *sk = sock->sk;
|
|
|
|
|
2006-03-21 07:48:35 +01:00
|
|
|
if (sk->sk_prot->compat_setsockopt != NULL)
|
|
|
|
return sk->sk_prot->compat_setsockopt(sk, level, optname,
|
|
|
|
optval, optlen);
|
2006-03-21 07:45:21 +01:00
|
|
|
return sk->sk_prot->setsockopt(sk, level, optname, optval, optlen);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(compat_sock_common_setsockopt);
|
|
|
|
#endif
|
|
|
|
|
2005-04-17 00:20:36 +02:00
|
|
|
void sk_common_release(struct sock *sk)
|
|
|
|
{
|
|
|
|
if (sk->sk_prot->destroy)
|
|
|
|
sk->sk_prot->destroy(sk);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Observation: when sock_common_release is called, processes have
|
|
|
|
* no access to socket. But net still has.
|
|
|
|
* Step one, detach it from networking:
|
|
|
|
*
|
|
|
|
* A. Remove from hash tables.
|
|
|
|
*/
|
|
|
|
|
|
|
|
sk->sk_prot->unhash(sk);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* In this point socket cannot receive new packets, but it is possible
|
|
|
|
* that some packets are in flight because some CPU runs receiver and
|
|
|
|
* did hash table lookup before we unhashed socket. They will achieve
|
|
|
|
* receive queue and will be purged by socket destructor.
|
|
|
|
*
|
|
|
|
* Also we still have packets pending on receive queue and probably,
|
|
|
|
* our own packets waiting in device queues. sock_destroy will drain
|
|
|
|
* receive queue, but transmitted packets will delay socket destruction
|
|
|
|
* until the last reference will be released.
|
|
|
|
*/
|
|
|
|
|
|
|
|
sock_orphan(sk);
|
|
|
|
|
|
|
|
xfrm_sk_free_policy(sk);
|
|
|
|
|
2005-08-10 04:45:38 +02:00
|
|
|
sk_refcnt_debug_release(sk);
|
2005-04-17 00:20:36 +02:00
|
|
|
sock_put(sk);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(sk_common_release);
|
|
|
|
|
|
|
|
static DEFINE_RWLOCK(proto_list_lock);
|
|
|
|
static LIST_HEAD(proto_list);
|
|
|
|
|
2008-03-29 00:38:17 +01:00
|
|
|
#ifdef CONFIG_PROC_FS
|
|
|
|
#define PROTO_INUSE_NR 64 /* should be enough for the first time */
|
2008-03-29 00:38:43 +01:00
|
|
|
struct prot_inuse {
|
|
|
|
int val[PROTO_INUSE_NR];
|
|
|
|
};
|
2008-03-29 00:38:17 +01:00
|
|
|
|
|
|
|
static DECLARE_BITMAP(proto_inuse_idx, PROTO_INUSE_NR);
|
2008-04-01 04:42:16 +02:00
|
|
|
|
|
|
|
#ifdef CONFIG_NET_NS
|
|
|
|
void sock_prot_inuse_add(struct net *net, struct proto *prot, int val)
|
|
|
|
{
|
|
|
|
int cpu = smp_processor_id();
|
|
|
|
per_cpu_ptr(net->core.inuse, cpu)->val[prot->inuse_idx] += val;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(sock_prot_inuse_add);
|
|
|
|
|
|
|
|
int sock_prot_inuse_get(struct net *net, struct proto *prot)
|
|
|
|
{
|
|
|
|
int cpu, idx = prot->inuse_idx;
|
|
|
|
int res = 0;
|
|
|
|
|
|
|
|
for_each_possible_cpu(cpu)
|
|
|
|
res += per_cpu_ptr(net->core.inuse, cpu)->val[idx];
|
|
|
|
|
|
|
|
return res >= 0 ? res : 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(sock_prot_inuse_get);
|
|
|
|
|
|
|
|
static int sock_inuse_init_net(struct net *net)
|
|
|
|
{
|
|
|
|
net->core.inuse = alloc_percpu(struct prot_inuse);
|
|
|
|
return net->core.inuse ? 0 : -ENOMEM;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void sock_inuse_exit_net(struct net *net)
|
|
|
|
{
|
|
|
|
free_percpu(net->core.inuse);
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct pernet_operations net_inuse_ops = {
|
|
|
|
.init = sock_inuse_init_net,
|
|
|
|
.exit = sock_inuse_exit_net,
|
|
|
|
};
|
|
|
|
|
|
|
|
static __init int net_inuse_init(void)
|
|
|
|
{
|
|
|
|
if (register_pernet_subsys(&net_inuse_ops))
|
|
|
|
panic("Cannot initialize net inuse counters");
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
core_initcall(net_inuse_init);
|
|
|
|
#else
|
2008-03-29 00:38:43 +01:00
|
|
|
static DEFINE_PER_CPU(struct prot_inuse, prot_inuse);
|
|
|
|
|
2008-04-01 04:41:46 +02:00
|
|
|
void sock_prot_inuse_add(struct net *net, struct proto *prot, int val)
|
2008-03-29 00:38:43 +01:00
|
|
|
{
|
|
|
|
__get_cpu_var(prot_inuse).val[prot->inuse_idx] += val;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(sock_prot_inuse_add);
|
|
|
|
|
2008-04-01 04:41:46 +02:00
|
|
|
int sock_prot_inuse_get(struct net *net, struct proto *prot)
|
2008-03-29 00:38:43 +01:00
|
|
|
{
|
|
|
|
int cpu, idx = prot->inuse_idx;
|
|
|
|
int res = 0;
|
|
|
|
|
|
|
|
for_each_possible_cpu(cpu)
|
|
|
|
res += per_cpu(prot_inuse, cpu).val[idx];
|
|
|
|
|
|
|
|
return res >= 0 ? res : 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(sock_prot_inuse_get);
|
2008-04-01 04:42:16 +02:00
|
|
|
#endif
|
2008-03-29 00:38:17 +01:00
|
|
|
|
|
|
|
static void assign_proto_idx(struct proto *prot)
|
|
|
|
{
|
|
|
|
prot->inuse_idx = find_first_zero_bit(proto_inuse_idx, PROTO_INUSE_NR);
|
|
|
|
|
|
|
|
if (unlikely(prot->inuse_idx == PROTO_INUSE_NR - 1)) {
|
|
|
|
printk(KERN_ERR "PROTO_INUSE_NR exhausted\n");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
set_bit(prot->inuse_idx, proto_inuse_idx);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void release_proto_idx(struct proto *prot)
|
|
|
|
{
|
|
|
|
if (prot->inuse_idx != PROTO_INUSE_NR - 1)
|
|
|
|
clear_bit(prot->inuse_idx, proto_inuse_idx);
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
static inline void assign_proto_idx(struct proto *prot)
|
|
|
|
{
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void release_proto_idx(struct proto *prot)
|
|
|
|
{
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2007-11-07 11:23:38 +01:00
|
|
|
int proto_register(struct proto *prot, int alloc_slab)
|
|
|
|
{
|
2005-04-17 00:20:36 +02:00
|
|
|
if (alloc_slab) {
|
|
|
|
prot->slab = kmem_cache_create(prot->name, prot->obj_size, 0,
|
udp: RCU handling for Unicast packets.
Goals are :
1) Optimizing handling of incoming Unicast UDP frames, so that no memory
writes should happen in the fast path.
Note: Multicasts and broadcasts still will need to take a lock,
because doing a full lockless lookup in this case is difficult.
2) No expensive operations in the socket bind/unhash phases :
- No expensive synchronize_rcu() calls.
- No added rcu_head in socket structure, increasing memory needs,
but more important, forcing us to use call_rcu() calls,
that have the bad property of making sockets structure cold.
(rcu grace period between socket freeing and its potential reuse
make this socket being cold in CPU cache).
David did a previous patch using call_rcu() and noticed a 20%
impact on TCP connection rates.
Quoting Cristopher Lameter :
"Right. That results in cacheline cooldown. You'd want to recycle
the object as they are cache hot on a per cpu basis. That is screwed
up by the delayed regular rcu processing. We have seen multiple
regressions due to cacheline cooldown.
The only choice in cacheline hot sensitive areas is to deal with the
complexity that comes with SLAB_DESTROY_BY_RCU or give up on RCU."
- Because udp sockets are allocated from dedicated kmem_cache,
use of SLAB_DESTROY_BY_RCU can help here.
Theory of operation :
---------------------
As the lookup is lockfree (using rcu_read_lock()/rcu_read_unlock()),
special attention must be taken by readers and writers.
Use of SLAB_DESTROY_BY_RCU is tricky too, because a socket can be freed,
reused, inserted in a different chain or in worst case in the same chain
while readers could do lookups in the same time.
In order to avoid loops, a reader must check each socket found in a chain
really belongs to the chain the reader was traversing. If it finds a
mismatch, lookup must start again at the begining. This *restart* loop
is the reason we had to use rdlock for the multicast case, because
we dont want to send same message several times to the same socket.
We use RCU only for fast path.
Thus, /proc/net/udp still takes spinlocks.
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2008-10-29 10:11:14 +01:00
|
|
|
SLAB_HWCACHE_ALIGN | prot->slab_flags,
|
|
|
|
NULL);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (prot->slab == NULL) {
|
|
|
|
printk(KERN_CRIT "%s: Can't create sock SLAB cache!\n",
|
|
|
|
prot->name);
|
2008-03-29 00:39:10 +01:00
|
|
|
goto out;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 07:46:52 +02:00
|
|
|
|
|
|
|
if (prot->rsk_prot != NULL) {
|
|
|
|
static const char mask[] = "request_sock_%s";
|
|
|
|
|
2008-11-22 01:45:22 +01:00
|
|
|
prot->rsk_prot->slab_name = kmalloc(strlen(prot->name) + sizeof(mask) - 1, GFP_KERNEL);
|
|
|
|
if (prot->rsk_prot->slab_name == NULL)
|
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 07:46:52 +02:00
|
|
|
goto out_free_sock_slab;
|
|
|
|
|
2008-11-22 01:45:22 +01:00
|
|
|
sprintf(prot->rsk_prot->slab_name, mask, prot->name);
|
|
|
|
prot->rsk_prot->slab = kmem_cache_create(prot->rsk_prot->slab_name,
|
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 07:46:52 +02:00
|
|
|
prot->rsk_prot->obj_size, 0,
|
2007-07-20 03:11:58 +02:00
|
|
|
SLAB_HWCACHE_ALIGN, NULL);
|
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 07:46:52 +02:00
|
|
|
|
|
|
|
if (prot->rsk_prot->slab == NULL) {
|
|
|
|
printk(KERN_CRIT "%s: Can't create request sock SLAB cache!\n",
|
|
|
|
prot->name);
|
|
|
|
goto out_free_request_sock_slab_name;
|
|
|
|
}
|
|
|
|
}
|
2005-08-10 05:09:30 +02:00
|
|
|
|
2005-12-14 08:25:19 +01:00
|
|
|
if (prot->twsk_prot != NULL) {
|
2005-08-10 05:09:30 +02:00
|
|
|
static const char mask[] = "tw_sock_%s";
|
|
|
|
|
2008-11-22 01:45:22 +01:00
|
|
|
prot->twsk_prot->twsk_slab_name = kmalloc(strlen(prot->name) + sizeof(mask) - 1, GFP_KERNEL);
|
2005-08-10 05:09:30 +02:00
|
|
|
|
2008-11-22 01:45:22 +01:00
|
|
|
if (prot->twsk_prot->twsk_slab_name == NULL)
|
2005-08-10 05:09:30 +02:00
|
|
|
goto out_free_request_sock_slab;
|
|
|
|
|
2008-11-22 01:45:22 +01:00
|
|
|
sprintf(prot->twsk_prot->twsk_slab_name, mask, prot->name);
|
2005-12-14 08:25:19 +01:00
|
|
|
prot->twsk_prot->twsk_slab =
|
2008-11-22 01:45:22 +01:00
|
|
|
kmem_cache_create(prot->twsk_prot->twsk_slab_name,
|
2005-12-14 08:25:19 +01:00
|
|
|
prot->twsk_prot->twsk_obj_size,
|
2008-11-17 04:40:17 +01:00
|
|
|
0,
|
|
|
|
SLAB_HWCACHE_ALIGN |
|
|
|
|
prot->slab_flags,
|
2007-07-20 03:11:58 +02:00
|
|
|
NULL);
|
2005-12-14 08:25:19 +01:00
|
|
|
if (prot->twsk_prot->twsk_slab == NULL)
|
2005-08-10 05:09:30 +02:00
|
|
|
goto out_free_timewait_sock_slab_name;
|
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2005-04-17 00:24:09 +02:00
|
|
|
write_lock(&proto_list_lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
list_add(&prot->node, &proto_list);
|
2008-03-29 00:38:17 +01:00
|
|
|
assign_proto_idx(prot);
|
2005-04-17 00:20:36 +02:00
|
|
|
write_unlock(&proto_list_lock);
|
2007-11-07 11:23:38 +01:00
|
|
|
return 0;
|
|
|
|
|
2005-08-10 05:09:30 +02:00
|
|
|
out_free_timewait_sock_slab_name:
|
2008-11-22 01:45:22 +01:00
|
|
|
kfree(prot->twsk_prot->twsk_slab_name);
|
2005-08-10 05:09:30 +02:00
|
|
|
out_free_request_sock_slab:
|
|
|
|
if (prot->rsk_prot && prot->rsk_prot->slab) {
|
|
|
|
kmem_cache_destroy(prot->rsk_prot->slab);
|
|
|
|
prot->rsk_prot->slab = NULL;
|
|
|
|
}
|
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 07:46:52 +02:00
|
|
|
out_free_request_sock_slab_name:
|
2008-11-22 01:45:22 +01:00
|
|
|
kfree(prot->rsk_prot->slab_name);
|
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 07:46:52 +02:00
|
|
|
out_free_sock_slab:
|
|
|
|
kmem_cache_destroy(prot->slab);
|
|
|
|
prot->slab = NULL;
|
2007-11-07 11:23:38 +01:00
|
|
|
out:
|
|
|
|
return -ENOBUFS;
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(proto_register);
|
|
|
|
|
|
|
|
void proto_unregister(struct proto *prot)
|
|
|
|
{
|
|
|
|
write_lock(&proto_list_lock);
|
2008-03-29 00:38:17 +01:00
|
|
|
release_proto_idx(prot);
|
2005-09-07 04:47:50 +02:00
|
|
|
list_del(&prot->node);
|
|
|
|
write_unlock(&proto_list_lock);
|
2005-04-17 00:20:36 +02:00
|
|
|
|
|
|
|
if (prot->slab != NULL) {
|
|
|
|
kmem_cache_destroy(prot->slab);
|
|
|
|
prot->slab = NULL;
|
|
|
|
}
|
|
|
|
|
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 07:46:52 +02:00
|
|
|
if (prot->rsk_prot != NULL && prot->rsk_prot->slab != NULL) {
|
|
|
|
kmem_cache_destroy(prot->rsk_prot->slab);
|
2008-11-22 01:45:22 +01:00
|
|
|
kfree(prot->rsk_prot->slab_name);
|
[NET] Generalise TCP's struct open_request minisock infrastructure
Kept this first changeset minimal, without changing existing names to
ease peer review.
Basicaly tcp_openreq_alloc now receives the or_calltable, that in turn
has two new members:
->slab, that replaces tcp_openreq_cachep
->obj_size, to inform the size of the openreq descendant for
a specific protocol
The protocol specific fields in struct open_request were moved to a
class hierarchy, with the things that are common to all connection
oriented PF_INET protocols in struct inet_request_sock, the TCP ones
in tcp_request_sock, that is an inet_request_sock, that is an
open_request.
I.e. this uses the same approach used for the struct sock class
hierarchy, with sk_prot indicating if the protocol wants to use the
open_request infrastructure by filling in sk_prot->rsk_prot with an
or_calltable.
Results? Performance is improved and TCP v4 now uses only 64 bytes per
open request minisock, down from 96 without this patch :-)
Next changeset will rename some of the structs, fields and functions
mentioned above, struct or_calltable is way unclear, better name it
struct request_sock_ops, s/struct open_request/struct request_sock/g,
etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
2005-06-19 07:46:52 +02:00
|
|
|
prot->rsk_prot->slab = NULL;
|
|
|
|
}
|
|
|
|
|
2005-12-14 08:25:19 +01:00
|
|
|
if (prot->twsk_prot != NULL && prot->twsk_prot->twsk_slab != NULL) {
|
|
|
|
kmem_cache_destroy(prot->twsk_prot->twsk_slab);
|
2008-11-22 01:45:22 +01:00
|
|
|
kfree(prot->twsk_prot->twsk_slab_name);
|
2005-12-14 08:25:19 +01:00
|
|
|
prot->twsk_prot->twsk_slab = NULL;
|
2005-08-10 05:09:30 +02:00
|
|
|
}
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(proto_unregister);
|
|
|
|
|
|
|
|
#ifdef CONFIG_PROC_FS
|
|
|
|
static void *proto_seq_start(struct seq_file *seq, loff_t *pos)
|
2008-01-02 06:58:02 +01:00
|
|
|
__acquires(proto_list_lock)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
read_lock(&proto_list_lock);
|
2007-07-09 22:15:14 +02:00
|
|
|
return seq_list_start_head(&proto_list, *pos);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
static void *proto_seq_next(struct seq_file *seq, void *v, loff_t *pos)
|
|
|
|
{
|
2007-07-09 22:15:14 +02:00
|
|
|
return seq_list_next(v, &proto_list, pos);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
static void proto_seq_stop(struct seq_file *seq, void *v)
|
2008-01-02 06:58:02 +01:00
|
|
|
__releases(proto_list_lock)
|
2005-04-17 00:20:36 +02:00
|
|
|
{
|
|
|
|
read_unlock(&proto_list_lock);
|
|
|
|
}
|
|
|
|
|
|
|
|
static char proto_method_implemented(const void *method)
|
|
|
|
{
|
|
|
|
return method == NULL ? 'n' : 'y';
|
|
|
|
}
|
|
|
|
|
|
|
|
static void proto_seq_printf(struct seq_file *seq, struct proto *proto)
|
|
|
|
{
|
|
|
|
seq_printf(seq, "%-9s %4u %6d %6d %-3s %6u %-3s %-10s "
|
|
|
|
"%2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c %2c\n",
|
|
|
|
proto->name,
|
|
|
|
proto->obj_size,
|
2008-11-20 00:14:01 +01:00
|
|
|
sock_prot_inuse_get(seq_file_net(seq), proto),
|
2005-04-17 00:20:36 +02:00
|
|
|
proto->memory_allocated != NULL ? atomic_read(proto->memory_allocated) : -1,
|
|
|
|
proto->memory_pressure != NULL ? *proto->memory_pressure ? "yes" : "no" : "NI",
|
|
|
|
proto->max_header,
|
|
|
|
proto->slab == NULL ? "no" : "yes",
|
|
|
|
module_name(proto->owner),
|
|
|
|
proto_method_implemented(proto->close),
|
|
|
|
proto_method_implemented(proto->connect),
|
|
|
|
proto_method_implemented(proto->disconnect),
|
|
|
|
proto_method_implemented(proto->accept),
|
|
|
|
proto_method_implemented(proto->ioctl),
|
|
|
|
proto_method_implemented(proto->init),
|
|
|
|
proto_method_implemented(proto->destroy),
|
|
|
|
proto_method_implemented(proto->shutdown),
|
|
|
|
proto_method_implemented(proto->setsockopt),
|
|
|
|
proto_method_implemented(proto->getsockopt),
|
|
|
|
proto_method_implemented(proto->sendmsg),
|
|
|
|
proto_method_implemented(proto->recvmsg),
|
|
|
|
proto_method_implemented(proto->sendpage),
|
|
|
|
proto_method_implemented(proto->bind),
|
|
|
|
proto_method_implemented(proto->backlog_rcv),
|
|
|
|
proto_method_implemented(proto->hash),
|
|
|
|
proto_method_implemented(proto->unhash),
|
|
|
|
proto_method_implemented(proto->get_port),
|
|
|
|
proto_method_implemented(proto->enter_memory_pressure));
|
|
|
|
}
|
|
|
|
|
|
|
|
static int proto_seq_show(struct seq_file *seq, void *v)
|
|
|
|
{
|
2007-07-09 22:15:14 +02:00
|
|
|
if (v == &proto_list)
|
2005-04-17 00:20:36 +02:00
|
|
|
seq_printf(seq, "%-9s %-4s %-8s %-6s %-5s %-7s %-4s %-10s %s",
|
|
|
|
"protocol",
|
|
|
|
"size",
|
|
|
|
"sockets",
|
|
|
|
"memory",
|
|
|
|
"press",
|
|
|
|
"maxhdr",
|
|
|
|
"slab",
|
|
|
|
"module",
|
|
|
|
"cl co di ac io in de sh ss gs se re sp bi br ha uh gp em\n");
|
|
|
|
else
|
2007-07-09 22:15:14 +02:00
|
|
|
proto_seq_printf(seq, list_entry(v, struct proto, node));
|
2005-04-17 00:20:36 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-03-12 22:34:29 +01:00
|
|
|
static const struct seq_operations proto_seq_ops = {
|
2005-04-17 00:20:36 +02:00
|
|
|
.start = proto_seq_start,
|
|
|
|
.next = proto_seq_next,
|
|
|
|
.stop = proto_seq_stop,
|
|
|
|
.show = proto_seq_show,
|
|
|
|
};
|
|
|
|
|
|
|
|
static int proto_seq_open(struct inode *inode, struct file *file)
|
|
|
|
{
|
2008-11-20 00:14:01 +01:00
|
|
|
return seq_open_net(inode, file, &proto_seq_ops,
|
|
|
|
sizeof(struct seq_net_private));
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
2007-02-12 09:55:35 +01:00
|
|
|
static const struct file_operations proto_seq_fops = {
|
2005-04-17 00:20:36 +02:00
|
|
|
.owner = THIS_MODULE,
|
|
|
|
.open = proto_seq_open,
|
|
|
|
.read = seq_read,
|
|
|
|
.llseek = seq_lseek,
|
2008-11-20 00:14:01 +01:00
|
|
|
.release = seq_release_net,
|
|
|
|
};
|
|
|
|
|
|
|
|
static __net_init int proto_init_net(struct net *net)
|
|
|
|
{
|
|
|
|
if (!proc_net_fops_create(net, "protocols", S_IRUGO, &proto_seq_fops))
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static __net_exit void proto_exit_net(struct net *net)
|
|
|
|
{
|
|
|
|
proc_net_remove(net, "protocols");
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static __net_initdata struct pernet_operations proto_net_ops = {
|
|
|
|
.init = proto_init_net,
|
|
|
|
.exit = proto_exit_net,
|
2005-04-17 00:20:36 +02:00
|
|
|
};
|
|
|
|
|
|
|
|
static int __init proto_init(void)
|
|
|
|
{
|
2008-11-20 00:14:01 +01:00
|
|
|
return register_pernet_subsys(&proto_net_ops);
|
2005-04-17 00:20:36 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
subsys_initcall(proto_init);
|
|
|
|
|
|
|
|
#endif /* PROC_FS */
|