kernel.git/include/net/route.h, branch linux-2.6.25.y

[IPV4] route cache: Introduce rt_genid for smooth cache invalidation

2008-02-01T03:28:27Z

Current ip route cache implementation is not suited to large caches. We can consume a lot of CPU when cache must be invalidated, since we currently need to evict all cache entries, and this eviction is sometimes asynchronous. min_delay & max_delay can somewhat control this asynchronism behavior, but whole thing is a kludge, regularly triggering infamous soft lockup messages. When entries are still in use, this also consumes a lot of ram, filling dst_garbage.list. A better scheme is to use a generation identifier on each entry, so that cache invalidation can be performed by changing the table identifier, without having to scan all entries. No more delayed flushing, no more stalling when secret_interval expires. Invalidated entries will then be freed at GC time (controled by ip_rt_gc_timeout or stress), or when an invalidated entry is found in a chain when an insert is done. Thus we keep a normal equilibrium. This patch : - renames rt_hash_rnd to rt_genid (and makes it an atomic_t) - Adds a new rt_genid field to 'struct rtable' (filling a hole on 64bit) - Checks entry->rt_genid at appropriate places :

[NET]: Introducing socket mark socket option.

2008-02-01T03:27:19Z

A userspace program may wish to set the mark for each packets its send without using the netfilter MARK target. Changing the mark can be used for mark based routing without netfilter or for packet filtering. It requires CAP_NET_ADMIN capability. Signed-off-by: Laszlo Attila Toth Acked-by: Patrick McHardy Signed-off-by: David S. Miller

[NETNS]: Routing cache virtualization.

2008-01-28T23:11:13Z

Basically, this piece looks relatively easy. Namespace is already available on the dst entry via device and the device is safe to dereferrence. Compare it with one of a searcher and skip entry if appropriate. The only exception is ip_rt_frag_needed. So, add namespace parameter to it. Signed-off-by: Denis V. Lunev Signed-off-by: David S. Miller

[NETNS]: Correct namespace for connect-time routing.

2008-01-28T23:11:12Z

ip_route_connect and ip_route_newports are a part of routing API presented to the socket layer. The namespace is available inside them through a socket. Signed-off-by: Denis V. Lunev Signed-off-by: David S. Miller

[NETNS]: Add namespace parameter to ip_route_output_key.

2008-01-28T23:11:07Z

Needed to propagate it down to the ip_route_output_flow. Signed-off-by: Denis V. Lunev Signed-off-by: David S. Miller

[NETNS]: Add namespace parameter to ip_route_output_flow.

2008-01-28T23:11:06Z

Needed to propagate it down to the __ip_route_output_key. Signed_off_by: Denis V. Lunev Signed-off-by: David S. Miller

[NETNS]: Add namespace parameter to __ip_route_output_key.

2008-01-28T23:11:05Z

This is only required to propagate it down to the ip_route_output_slow. Signed-off-by: Denis V. Lunev Signed-off-by: David S. Miller

[NETNS]: Pass namespace through ip_rt_ioctl.

2008-01-28T23:01:34Z

... up to rtentry_to_fib_config Acked-by: Benjamin Thery Acked-by: Daniel Lezcano Signed-off-by: Denis V. Lunev Signed-off-by: David S. Miller

[NETNS]: Add netns parameter to inet_(dev_)add_type.

2008-01-28T23:01:27Z

The patch extends the inet_addr_type and inet_dev_addr_type with the network namespace pointer. That allows to access the different tables relatively to the network namespace. The modification of the signature function is reported in all the callers of the inet_addr_type using the pointer to the well known init_net. Acked-by: Benjamin Thery Acked-by: Daniel Lezcano Signed-off-by: Eric W. Biederman Signed-off-by: David S. Miller

[IPV4]: Add inet_dev_addr_type()

2008-01-28T22:55:56Z

Address type search can be limited to an interface by inet_dev_addr_type function. Signed-off-by: Laszlo Attila Toth Signed-off-by: Patrick McHardy Signed-off-by: David S. Miller