]> git.proxmox.com Git - mirror_ubuntu-disco-kernel.git/commit
[XFRM]: Allow packet drops during larval state resolution.
authorDavid S. Miller <davem@sunset.davemloft.net>
Fri, 25 May 2007 01:17:54 +0000 (18:17 -0700)
committerDavid S. Miller <davem@sunset.davemloft.net>
Fri, 25 May 2007 01:17:54 +0000 (18:17 -0700)
commit14e50e57aedb2a89cf79b77782879769794cab7b
tree46cbdab9c8007cea0821294c9d397214b38ea4c8
parent04efb8787e4d8a7b21a61aeb723de33154311256
[XFRM]: Allow packet drops during larval state resolution.

The current IPSEC rule resolution behavior we have does not work for a
lot of people, even though technically it's an improvement from the
-EAGAIN buisness we had before.

Right now we'll block until the key manager resolves the route.  That
works for simple cases, but many folks would rather packets get
silently dropped until the key manager resolves the IPSEC rules.

We can't tell these folks to "set the socket non-blocking" because
they don't have control over the non-block setting of things like the
sockets used to resolve DNS deep inside of the resolver libraries in
libc.

With that in mind I coded up the patch below with some help from
Herbert Xu which provides packet-drop behavior during larval state
resolution, controllable via sysctl and off by default.

This lays the framework to either:

1) Make this default at some point or...

2) Move this logic into xfrm{4,6}_policy.c and implement the
   ARP-like resolution queue we've all been dreaming of.
   The idea would be to queue packets to the policy, then
   once the larval state is resolved by the key manager we
   re-resolve the route and push the packets out.  The
   packets would timeout if the rule didn't get resolved
   in a certain amount of time.

Signed-off-by: David S. Miller <davem@davemloft.net>
include/net/dst.h
include/net/ipv6.h
net/core/sysctl_net_core.c
net/dccp/ipv6.c
net/ipv4/route.c
net/ipv6/datagram.c
net/ipv6/raw.c
net/ipv6/route.c
net/ipv6/tcp_ipv6.c
net/ipv6/udp.c
net/xfrm/xfrm_policy.c