Allan Stephens [Wed, 16 Apr 2008 02:02:30 +0000 (19:02 -0700)]
[TIPC]: Use fast buffer cloning to improve performance
This patch causes TIPC to allocate fast clonable sk_buffs,
rather than standard ones. This speeds up the cloning
operation done by the link code each time a message is sent
off-node.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Tue, 15 Apr 2008 07:36:08 +0000 (00:36 -0700)]
[NETNS]: The generic per-net pointers.
Add the elastic array of void * pointer to the struct net.
The access rules are simple:
1. register the ops with register_pernet_gen_device to get
the id of your private pointer
2. call net_assign_generic() to put the private data on the
struct net (most preferably this should be done in the
->init callback of the ops registered)
3. do not store any private reference on the net_generic array;
4. do not change this pointer while the net is alive;
5. use the net_generic() to get the pointer.
When adding a new pointer, I copy the old array, replace it
with a new one and schedule the old for kfree after an RCU
grace period.
Since the net_generic explores the net->gen array inside rcu
read section and once set the net->gen->ptr[x] pointer never
changes, this grants us a safe access to generic pointers.
Quoting Paul: "... RCU is protecting -only- the net_generic
structure that net_generic() is traversing, and the [pointer]
returned by net_generic() is protected by a reference counter
in the upper-level struct net."
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Tue, 15 Apr 2008 07:35:23 +0000 (00:35 -0700)]
[NETNS]: The net-subsys IDs generator.
To make some per-net generic pointers, we need some way to address
them, i.e. - IDs. This is simple IDA-based IDs generator for pernet
subsystems.
Addressing questions about potential checkpoint/restart problems:
these IDs are "lite-offsets" within the net structure and are by no
means supposed to be exported to the userspace.
Since it will be used in the nearest future by devices only (tun,
vlan, tunnels, bridge, etc), I make it resemble the functionality
of register_pernet_device().
The new ids is stored in the *id pointer _before_ calling the init
callback to make this id available in this callback.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Adrian Bunk [Tue, 15 Apr 2008 07:30:16 +0000 (00:30 -0700)]
[ISDN] include/linux/isdn.h: remove dead code
This patch remove the usage of a nonexisting kconfig variable.
Reported-by: Robert P. J. Day <rpjday@crashcourse.ca> Signed-off-by: Adrian Bunk <bunk@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Adrian Bunk [Tue, 15 Apr 2008 07:29:24 +0000 (00:29 -0700)]
[IRDA]: Remove irlan_eth_send_gratuitous_arp()
Even kernel 2.2.26 (sic) already contains the
#undef CONFIG_IRLAN_SEND_GRATUITOUS_ARP
with the comment "but for some reason the machine crashes if you use DHCP".
Either someone finally looks into this or it's simply time to remove
this dead code.
Reported-by: Robert P. J. Day <rpjday@crashcourse.ca> Signed-off-by: Adrian Bunk <bunk@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Allan Stephens [Tue, 15 Apr 2008 07:22:02 +0000 (00:22 -0700)]
[TIPC]: Overhaul of socket locking logic
This patch modifies TIPC's socket code to follow the same approach
used by other protocols. This change eliminates the need for a
mutex in the TIPC-specific portion of the socket protocol data
structure -- in its place, the standard Linux socket backlog queue
and associated locking routines are utilized. These changes fix
a long-standing receive queue bug on SMP systems, and also enable
individual read and write threads to utilize a socket without
unnecessarily interfering with each other.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Allan Stephens [Tue, 15 Apr 2008 07:15:50 +0000 (00:15 -0700)]
[TIPC]: Correct "off by 1" error in socket queue limit enforcement
This patch fixes a bug that allowed TIPC to queue 1 more message
than allowed by the socket receive queue threshold limits. The
patch also improves the threshold code's logic and naming to help
prevent this sort of error from recurring in the future.
Signed-off-by: Allan Stephens <allan.stephens@windriver.com> Signed-off-by: David S. Miller <davem@davemloft.net>
When CONFIG_SECURITY_NETWORK_XFRM is undefined the following warnings appears:
net/xfrm/xfrm_user.c: In function 'xfrm_add_pol_expire':
net/xfrm/xfrm_user.c:1576: warning: 'ctx' may be used uninitialized in this function
net/xfrm/xfrm_user.c: In function 'xfrm_get_policy':
net/xfrm/xfrm_user.c:1340: warning: 'ctx' may be used uninitialized in this function
(security_xfrm_policy_alloc is noop for the case).
Patrick McHardy [Mon, 14 Apr 2008 09:15:52 +0000 (11:15 +0200)]
[NETFILTER]: nf_nat: kill helper and seq_adjust hooks
Connection tracking helpers (specifically FTP) need to be called
before NAT sequence numbers adjustments are performed to be able
to compare them against previously seen ones. We've introduced
two new hooks around 2.6.11 to maintain this ordering when NAT
modules were changed to get called from conntrack helpers directly.
The cost of netfilter hooks is quite high and sequence number
adjustments are only rarely needed however. Add a RCU-protected
sequence number adjustment function pointer and call it from
IPv4 conntrack after calling the helper.
Patrick McHardy [Mon, 14 Apr 2008 09:15:51 +0000 (11:15 +0200)]
[NETFILTER]: nf_nat: don't add NAT extension for confirmed conntracks
Adding extensions to confirmed conntracks is not allowed to avoid races
on reallocation. Don't setup NAT for confirmed conntracks in case NAT
module is loaded late.
The has one side-effect, the connections existing before the NAT module
was loaded won't enter the bysource hash. The only case where this actually
makes a difference is in case of SNAT to a multirange where the IP before
NAT is also part of the range. Since old connections don't enter the
bysource hash the first new connection from the IP will have a new address
selected. This shouldn't matter at all.
Patrick McHardy [Mon, 14 Apr 2008 09:15:50 +0000 (11:15 +0200)]
[NETFILTER]: nf_nat: remove obsolete check for ICMP redirects
Locally generated ICMP packets have a reference to the conntrack entry
of the original packet manually attached by icmp_send(). Therefore the
check for locally originated untracked ICMP redirects can never be
true.
Move the UDP-Lite conntrack checksum validation to a generic helper
similar to nf_checksum() and make it fall back to nf_checksum()
in case the full packet is to be checksummed and hardware checksums
are available. This is to be used by DCCP conntrack, which also
needs to verify partial checksums.
Patrick McHardy [Mon, 14 Apr 2008 09:15:47 +0000 (11:15 +0200)]
[NETFILTER]: nf_conntrack_netlink: clean up NAT protocol parsing
Move responsibility for setting the IP_NAT_RANGE_PROTO_SPECIFIED flag
to the NAT protocol, properly propagate errors and get rid of ugly
return value convention.
Patrick McHardy [Mon, 14 Apr 2008 09:15:45 +0000 (11:15 +0200)]
[NETFILTER]: {ip,ip6,arp}_tables: return EAGAIN for invalid SO_GET_ENTRIES size
Rule dumping is performed in two steps: first userspace gets the
ruleset size using getsockopt(SO_GET_INFO) and allocates memory,
then it calls getsockopt(SO_GET_ENTRIES) to actually dump the
ruleset. When another process changes the ruleset in between the
sizes from the first getsockopt call doesn't match anymore and
the kernel aborts. Unfortunately it returns EAGAIN, as for multiple
other possible errors, so userspace can't distinguish this case
from real errors.
Return EAGAIN so userspace can retry the operation.
Fixes (with current iptables SVN version) netfilter bugzilla #104.
Jan Engelhardt [Mon, 14 Apr 2008 07:56:04 +0000 (09:56 +0200)]
[NETFILTER]: xt_sctp: simplify xt_sctp.h
The use of xt_sctp.h flagged up -Wshadow warnings in userspace, which
prompted me to look at it and clean it up. Basic operations have been
directly replaced by library calls (memcpy, memset is both available
in the kernel and userspace, and usually faster than a self-made
loop). The is_set and is_clear functions now use a processing time
shortcut, too.
Signed-off-by: Jan Engelhardt <jengelh@computergmbh.de> Signed-off-by: Patrick McHardy <kaber@trash.net>
[NETFILTER]: ip_tables: per-netns FILTER/MANGLE/RAW tables for real
Commit 9335f047fe61587ec82ff12fbb1220bcfdd32006 aka
"[NETFILTER]: ip_tables: per-netns FILTER, MANGLE, RAW"
added per-netns _view_ of iptables rules. They were shown to user, but
ignored by filtering code. Now that it's possible to at least ping loopback,
per-netns tables can affect filtering decisions.
netns is taken in case of
PRE_ROUTING, LOCAL_IN -- from in device,
POST_ROUTING, LOCAL_OUT -- from out device,
FORWARD -- from in device which should be equal to out device's netns.
This code is relatively new, so BUG_ON was plugged.
Wrappers were added to a) keep code the same from CONFIG_NET_NS=n users
(overwhelming majority), b) consolidate code in one place -- similar
changes will be done in ipv6 and arp netfilter code.
Signed-off-by: Alexey Dobriyan <adobriyan@sw.ru> Signed-off-by: Patrick McHardy <kaber@trash.net>
Patrick McHardy [Thu, 20 Mar 2008 14:15:45 +0000 (15:15 +0100)]
[NETFILTER]: {ip,ip6}t_LOG: print MARK value in log output
Dump the mark value in log messages similar to nfnetlink_log. This
is useful for debugging complex setups where marks are used for
routing or traffic classification.
Alexey Dobriyan [Thu, 20 Mar 2008 14:15:43 +0000 (15:15 +0100)]
[NETFILTER]: nf_conntrack: less hairy ifdefs around proc and sysctl
Patch splits creation of /proc/net/nf_conntrack, /proc/net/stat/nf_conntrack
and net.netfilter hierarchy into their own functions with dummy ones
if PROC_FS or SYSCTL is not set. Also, remove dead "ret = 0" write
while I'm at it.
Signed-off-by: Alexey Dobriyan <adobriyan@sw.ru> Signed-off-by: Patrick McHardy <kaber@trash.net>
Patrick McHardy [Mon, 14 Apr 2008 07:46:01 +0000 (00:46 -0700)]
[BRIDGE]: Fix crash in __ip_route_output_key with bridge netfilter
The bridge netfilter code attaches a fake dst_entry with a pointer to a
fake net_device structure to skbs it passes up to IPv4 netfilter. This
leads to crashes when the skb is passed to __ip_route_output_key when
dereferencing the namespace pointer.
Since bridging can currently only operate in the init_net namespace,
the easiest fix for now is to initialize the nd_net pointer of the
fake net_device struct to &init_net.
Should fix bugzilla 10323: http://bugzilla.kernel.org/show_bug.cgi?id=10323
Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 07:44:52 +0000 (00:44 -0700)]
[NETFILTER]: ipt_CLUSTERIP: fix race between clusterip_config_find_get and _entry_put
Consider we are putting a clusterip_config entry with the "entries"
count == 1, and on the other CPU there's a clusterip_config_find_get
in progress:
CPU1: CPU2:
clusterip_config_entry_put: clusterip_config_find_get:
if (atomic_dec_and_test(&c->entries)) {
/* true */
read_lock_bh(&clusterip_lock);
c = __clusterip_config_find(clusterip);
/* found - it's still in list */
...
atomic_inc(&c->entries);
read_unlock_bh(&clusterip_lock);
Oops! We have an entry returned by the clusterip_config_find_get,
which is a) not in list b) has a stale dev pointer.
The problems will happen when the CPU2 will release the entry - it
will remove it from the list for the 2nd time, thus spoiling it, and
will put a stale dev pointer.
The fix is to make atomic_dec_and_test under the clusterip_lock.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Signed-off-by: Patrick McHardy <kaber@trash.net>
[IPV6] MROUTE: Add stats in multicast routing module method ip6_mr_forward().
This patches adds a call to increment IPSTATS_MIB_OUTFORWDATAGRAMS
when forwarding the packet in ip6_mr_forward() in the IPv6 multicast
routing module (net/ipv6/ip6mr.c).
Signed-off-by: Rami Rosen <ramirose@gmail.com> Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
[IPV6]: IPv6 extension header structures need to be packed.
struct ipv6_opt_hdr is the common structure for IPv6 extension
headers, and it is common to increment the pointer to get
the real content. On the other hand, since the structure
consists only of 1-byte next-header field and 1-byte length
field, size of that structure depends on architecture; 2 or 4.
Add "packed" attribute to get 2.
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Patrick McHardy [Mon, 14 Apr 2008 05:45:40 +0000 (22:45 -0700)]
[NET]: Return more appropriate error from eth_validate_addr().
Paul Bolle wrote:
> http://bugzilla.kernel.org/show_bug.cgi?id=9923 would have been much easier to
> track down if eth_validate_addr() would somehow complain aloud if an address
> is invalid. Shouldn't it make at least some noise?
I guess it should return -EADDRNOTAVAIL similar to eth_mac_addr()
when validation fails.
Signed-off-by: Patrick McHardy <kaber@trash.net> Signed-off-by: David S. Miller <davem@davemloft.net>
Paul Bolle [Mon, 14 Apr 2008 05:44:20 +0000 (22:44 -0700)]
[ISDN]: Do not validate ISDN net device address prior to interface-up
Commit bada339 (Validate device addr prior to interface-up) caused a regression
in the ISDN network code, see: http://bugzilla.kernel.org/show_bug.cgi?id=9923
The trivial fix is to remove the pointer to eth_validate_addr() in the
net_device struct in isdn_net_init().
Signed-off-by: Paul Bolle <pebolle@tiscali.nl> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 05:32:45 +0000 (22:32 -0700)]
[NETNS][DCCPV6]: Actually create ctl socket on each net and use it.
Move the call to inet_ctl_sock_create to init callback (and
inet_ctl_sock_destroy to exit one) and use proper ctl sock
in dccp_v6_ctl_send_reset.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 05:32:25 +0000 (22:32 -0700)]
[NETNS][DCCPV6]: Move the dccp_v6_ctl_sk on the struct net.
And replace all its usage with init_net's socket.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 05:32:02 +0000 (22:32 -0700)]
[NETNS][DCCPV6]: Add dummy per-net operations.
They will be responsible for ctl socket initialization, but
currently they are void.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 05:31:32 +0000 (22:31 -0700)]
[NETNS][DCCPV6]: Don't pass NULL to ip6_dst_lookup.
This call uses the sock to get the net to lookup the routing
in. With CONFIG_NET_NS this code will OOPS, since the sk ptr
is NULL.
After looking inside the ip6_dst_lookup and drawing the analogy
with respective ipv6 code, it seems, that the dccp ctl socket
is a good candidate for the first argument.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 05:31:05 +0000 (22:31 -0700)]
[NETNS][DCCPV4]: Enable DCCPv4 in net namespaces.
This enables sockets creation with IPPROTO_DCCP and enables
the ip level to pass DCCP packets to the DCCP level.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 05:30:43 +0000 (22:30 -0700)]
[NETNS][DCCPV4]: Make per-net socket lookup.
The inet_lookup family of functions requires a net to lookup
a socket in, so give a proper one to them.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 05:30:19 +0000 (22:30 -0700)]
[NETNS][DCCPV4]: Use proper net to route the reset packet.
The dccp_v4_route_skb used in dccp_v4_ctl_send_reset, currently
works with init_net's routing tables - fix it.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 05:29:59 +0000 (22:29 -0700)]
[NETNS][DCCPV4]: Actually create ctl socket on each net and use it.
Move the call to inet_ctl_sock_create to init callback (and
inet_ctl_sock_destroy to exit one) and use proper ctl sock
in dccp_v4_ctl_send_reset.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 05:29:37 +0000 (22:29 -0700)]
[NETNS][DCCPV4]: Move the dccp_v4_ctl_sk on the struct net.
And replace all its usage with init_net's socket.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 05:29:13 +0000 (22:29 -0700)]
[NETNS][DCCPV4]: Add dummy per-net operations.
They will be responsible for ctl socket initialization, but
currently they are void.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Mon, 14 Apr 2008 05:28:42 +0000 (22:28 -0700)]
[NETNS]: Add an empty netns_dccp structure on struct net.
According to the overall struct net design, it will be
filled with DCCP-related members.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/atm/horizon.c has unusually large number
of static inline functions - 36.
I looked through them. Most of them seems to be small enough,
but a few are big, others are using udelay or busy loop,
and as such are better not be inlined.
This patch removes "inline" from these static functions
(regardless of number of callsites - gcc nowadays auto-inlines
statics with one callsite).
Size difference for 32bit x86:
text data bss dec hex filename
8201 180 6 8387 20c3 linux-2.6-ALLYES/drivers/atm/horizon.o
7840 180 6 8026 1f5a linux-2.6.inline-ALLYES/drivers/atm/horizon.o
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
sk_reuse is declared as "unsigned char", but is set as type valbool in net/core/sock.c.
There is no other place in net/ where sk->sk_reuse is set to a value > 1, so the test
"sk_reuse > 1" can not be true.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / linux-2.6 / log