Merge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
[TCP]: Fix minisock tcp_create_openreq_child() typo.
[TCP]: Document several sysctls.
[NET]: Fix kfree(skb)
[NET]: Handle disabled preemption in gfp_any()
[BRIDGE]: Fix locking of set path cost.
[IPV6]: /proc/net/anycast6 unbalanced inet6_dev refcnt
[IPX]: Remove ancient changelog
[IPX]: Remove outdated information from Kconfig
[NET]: Revert socket.h/stat.h ifdef hacks.
[IPV6]: anycast refcnt fix
[XFRM] xfrm_user: Fix return values of xfrm_add_sa_expire.

Documentation/networking/ip-sysctl.txt

···
 	More congestion control algorithms may be available as modules,
 	but not loaded.
 
+tcp_base_mss - INTEGER
+	The initial value of search_low to be used by Packetization Layer
+	Path MTU Discovery (MTU probing).  If MTU probing is enabled,
+	this is the inital MSS used by the connection.
+
 tcp_congestion_control - STRING
 	Set the congestion control algorithm to be used for new
 	connections. The algorithm "reno" is always available, but
···
 
 	Defaults are calculated at boot time from amount of available
 	memory.
+
+tcp_moderate_rcvbuf - BOOLEAN
+	If set, TCP performs receive buffer autotuning, attempting to
+	automatically size the buffer (no greater than tcp_rmem[2]) to
+	match the size required by the path for full throughput.  Enabled by
+	default.
+
+tcp_mtu_probing - INTEGER
+	Controls TCP Packetization-Layer Path MTU Discovery.  Takes three
+	values:
+	  0 - Disabled
+	  1 - Disabled by default, enabled when an ICMP black hole detected
+	  2 - Always enabled, use initial MSS of tcp_base_mss.
+
+tcp_no_metrics_save - BOOLEAN
+	By default, TCP saves various connection metrics in the route cache
+	when the connection closes, so that connections established in the
+	near future can use these to set initial conditions.  Usually, this
+	increases overall performance, but may sometimes cause performance
+	degredation.  If set, TCP will not cache metrics on closing
+	connections.
 
 tcp_orphan_retries - INTEGER
 	How may times to retry before killing TCP connection, closed

include/linux/socket.h

···
 				/* _SS_MAXSIZE value minus size of ss_family */
 } __attribute__ ((aligned(_K_SS_ALIGNSIZE)));	/* force desired alignment */
 
-#ifdef __KERNEL__
+#if defined(__KERNEL__) || !defined(__GLIBC__) || (__GLIBC__ < 2)
 
 #include <asm/socket.h>			/* arch-dependent defines	*/
 #include <linux/sockios.h>		/* the SIOCxxx I/O controls	*/

include/linux/stat.h

···
 
 #endif
 
-#ifdef __KERNEL__
+#if defined(__KERNEL__) || !defined(__GLIBC__) || (__GLIBC__ < 2)
 
 #define S_IFMT  00170000
 #define S_IFSOCK 0140000

include/net/sock.h

···
 
 static inline gfp_t gfp_any(void)
 {
-	return in_softirq() ? GFP_ATOMIC : GFP_KERNEL;
+	return in_atomic() ? GFP_ATOMIC : GFP_KERNEL;
 }
 
 static inline long sock_rcvtimeo(const struct sock *sk, int noblock)

net/bridge/br_ioctl.c

···
 		if (!capable(CAP_NET_ADMIN))
 			return -EPERM;
 
-		spin_lock_bh(&br->lock);
 		if ((p = br_get_port(br, args[1])) == NULL)
 			ret = -EINVAL;
 		else
 			br_stp_set_path_cost(p, args[2]);
-		spin_unlock_bh(&br->lock);
+
 		return ret;
 	}
 

net/core/skbuff.c

···
 err:
 	while ((skb = segs)) {
 		segs = skb->next;
-		kfree(skb);
+		kfree_skb(skb);
 	}
 	return ERR_PTR(err);
 }

net/dccp/output.c

···
 					 err);
 		} else {
 			dccp_pr_debug("packet discarded\n");
-			kfree(skb);
+			kfree_skb(skb);
 		}
 	}
 }

net/ipv4/tcp_minisocks.c

···
 	if (newsk != NULL) {
 		const struct inet_request_sock *ireq = inet_rsk(req);
 		struct tcp_request_sock *treq = tcp_rsk(req);
-		struct inet_connection_sock *newicsk = inet_csk(sk);
+		struct inet_connection_sock *newicsk = inet_csk(newsk);
 		struct tcp_sock *newtp;
 
 		/* Now setup tcp_sock */

net/ipv6/addrconf.c

···
 			ipv6_dev_mc_dec(dev, &addr);
 	}
 	for (ifa=idev->addr_list; ifa; ifa=ifa->if_next) {
+		if (ifa->flags&IFA_F_TENTATIVE)
+			continue;
 		if (idev->cnf.forwarding)
 			addrconf_join_anycast(ifa);
 		else

net/ipv6/anycast.c

···
 			break;
 		}
 		read_unlock_bh(&idev->lock);
+		in6_dev_put(idev);
 	}
 	return im;
 }

net/ipx/ChangeLog

···
- Revision 0.21:	Uses the new generic socket option code.
-
- Revision 0.22:	Gcc clean ups and drop out device registration. Use the
- 		new multi-protocol edition of hard_header
-
- Revision 0.23: IPX /proc by Mark Evans. Adding a route will
- 		will overwrite any existing route to the same network.
-
- Revision 0.24:	Supports new /proc with no 4K limit
-
- Revision 0.25:	Add ephemeral sockets, passive local network
- 		identification, support for local net 0 and
- 		multiple datalinks <Greg Page>
-
- Revision 0.26: Device drop kills IPX routes via it. (needed for module)
-
- Revision 0.27: Autobind <Mark Evans>
-
- Revision 0.28: Small fix for multiple local networks <Thomas Winder>
-
- Revision 0.29: Assorted major errors removed <Mark Evans>
- 		Small correction to promisc mode error fix <Alan Cox>
- 		Asynchronous I/O support. Changed to use notifiers
- 		and the newer packet_type stuff. Assorted major
- 		fixes <Alejandro Liu>
-
- Revision 0.30:	Moved to net/ipx/...	<Alan Cox>
- 		Don't set address length on recvfrom that errors.
- 		Incorrect verify_area.
-
- Revision 0.31:	New sk_buffs. This still needs a lot of 
- 		testing. <Alan Cox>
-
- Revision 0.32: Using sock_alloc_send_skb, firewall hooks. <Alan Cox>
- 		Supports sendmsg/recvmsg
-
- Revision 0.33:	Internal network support, routing changes, uses a
- 		protocol private area for ipx data.
-
- Revision 0.34:	Module support. <Jim Freeman>
-
- Revision 0.35: Checksum support. <Neil Turton>, hooked in by <Alan Cox>
- 		Handles WIN95 discovery packets <Volker Lendecke>
-
- Revision 0.36:	Internal bump up for 2.1
-
- Revision 0.37:	Began adding POSIXisms.
-
- Revision 0.38: Asynchronous socket stuff made current.
-
- Revision 0.39: SPX interfaces
-
- Revision 0.40: Tiny SIOCGSTAMP fix ([e-mail])
-
- Revision 0.41: 802.2TR removed ([e-mail])
- 		Fixed connecting to primary net,
- 		Automatic binding on send & receive,
- 		Martijn van Oosterhout <[e-mail]>
-
- Revision 042:  Multithreading - use spinlocks and refcounting to
- 		protect some structures: ipx_interface sock list, list
- 		of ipx interfaces, etc. 
- 		Bugfixes - do refcounting on net_devices, check function
- 		results, etc. Thanks to davem and freitag for
- 		suggestions and guidance.
- 		Arnaldo Carvalho de Melo <[e-mail]>,
- 		November, 2000
-
- Revision 043:	Shared SKBs, don't mangle packets, some cleanups
- 		Arnaldo Carvalho de Melo <[e-mail]>,
- 		December, 2000
-
- Revision 044:	Call ipxitf_hold on NETDEV_UP - acme
-
- Revision 045:	fix PPROP routing bug - acme
-
- Revision 046:	Further fixes to PPROP, ipxitf_create_internal was
- 		doing an unneeded MOD_INC_USE_COUNT, implement
- 		sysctl for ipx_pprop_broacasting, fix the ipx sysctl
- 		handling, making it dynamic, some cleanups, thanks to
- 		Petr Vandrovec for review and good suggestions. (acme)
-
- Revision 047:	Cleanups, CodingStyle changes, move the ncp connection
- 		hack out of line - acme
-
- Revision 048:	Use sk->protinfo to store the pointer to IPX private
- 		area, remove af_ipx from sk->protinfo and move ipx_opt
- 		to include/net/ipx.h, use IPX_SK like DecNET, etc - acme
-
- Revision 049:	SPX support dropped, see comment in ipx_create - acme
-
- Revision 050:	Use seq_file for proc stuff, moving it to ipx_proc.c - acme
-
-Other fixes:
- 
- Protect the module by a MOD_INC_USE_COUNT/MOD_DEC_USE_COUNT pair. Also, now
- usage count is managed this way:
- -Count one if the auto_interface mode is on
- -Count one per configured interface
- 
- Jacques Gelinas ([e-mail])

net/ipx/Kconfig

···
 	  support", below.
 
 	  IPX is similar in scope to IP, while SPX, which runs on top of IPX,
-	  is similar to TCP. There is also experimental support for SPX in
-	  Linux (see "SPX networking", below).
+	  is similar to TCP.
 
 	  To turn your Linux box into a fully featured NetWare file server and
 	  IPX router, say Y here and fetch either lwared from
···
 	  mars_nwe from <ftp://www.compu-art.de/mars_nwe/>. For more
 	  information, read the IPX-HOWTO available from
 	  <http://www.tldp.org/docs.html#howto>.
-
-	  General information about how to connect Linux, Windows machines and
-	  Macs is on the WWW at <http://www.eats.com/linux_mac_win.html>.
 
 	  The IPX driver would enlarge your kernel by about 16 KB. To compile
 	  this driver as a module, choose M here: the module will be called ipx.

net/netfilter/nfnetlink.c

···
 
 int nfnetlink_send(struct sk_buff *skb, u32 pid, unsigned group, int echo)
 {
-	gfp_t allocation = in_interrupt() ? GFP_ATOMIC : GFP_KERNEL;
 	int err = 0;
 
 	NETLINK_CB(skb).dst_group = group;
 	if (echo)
 		atomic_inc(&skb->users);
-	netlink_broadcast(nfnl, skb, pid, group, allocation);
+	netlink_broadcast(nfnl, skb, pid, group, gfp_any());
 	if (echo)
 		err = netlink_unicast(nfnl, skb, pid, MSG_DONTWAIT);
 

net/netlabel/netlabel_unlabeled.c

···
 	return 0;
 
 list_failure:
-	kfree(ans_skb);
+	kfree_skb(ans_skb);
 	return ret_val;
 }
 

net/xfrm/xfrm_user.c

···
 
 	x = xfrm_state_lookup(&id->daddr, id->spi, id->proto, id->family);
 	if (x == NULL) {
-		kfree(r_skb);
+		kfree_skb(r_skb);
 		return -ESRCH;
 	}
 
···
 	struct xfrm_usersa_info *p = &ue->state;
 
 	x = xfrm_state_lookup(&p->id.daddr, p->id.spi, p->id.proto, p->family);
-		err = -ENOENT;
 
+	err = -ENOENT;
 	if (x == NULL)
 		return err;
 
-	err = -EINVAL;
-
 	spin_lock_bh(&x->lock);
+	err = -EINVAL;
 	if (x->km.state != XFRM_STATE_VALID)
 		goto out;
 	km_state_expired(x, ue->hard, current->pid);
···
 		xfrm_audit_log(NETLINK_CB(skb).loginuid, NETLINK_CB(skb).sid,
 			       AUDIT_MAC_IPSEC_DELSA, 1, NULL, x);
 	}
+	err = 0;
 out:
 	spin_unlock_bh(&x->lock);
 	xfrm_state_put(x);