Re: [PATCH net-next] route: Add a new fib_multipath_hash_policy base on cpu id for tunnel packet

[Ido Schimmel <idosch@mellanox.com>]

On Fri, Feb 22, 2019 at 10:19:41PM -0500, David Ahern wrote:
> On 2/21/19 10:52 PM, wenxu@ucloud.cn wrote:
> > From: wenxu <wenxu@ucloud.cn>
> > 
> > Current fib_multipath_hash_policy can make hash based on the L3 or
> > L4. But it only work on the outer IP. So a specific tunnel always
> > has the same hash value. But a specific tunnel may contain so many
> > inner connection. However there is no good ways for tunnel packet.
> > A specific tunnel route based on the percpu dst_cache, It will not
> > lookup route table each packet.
> > 
> > This patch provide a based cpu id hash policy. The different
> > connection run on differnt cpu and There will differnet hash
> > value for percpu dst_cache.
> > 
> > Signed-off-by: wenxu <wenxu@ucloud.cn>
> > ---
> >  net/ipv4/route.c           | 6 ++++++
> >  net/ipv4/sysctl_net_ipv4.c | 2 +-
> >  2 files changed, 7 insertions(+), 1 deletion(-)
> > 
> 
> This multipath hash policy is global - for all fib lookups, not just
> tunnels.
> 
> The suggestion by Nik is worth exploring - an option to add the mark to
> the hash (e.g., L3 header + mark) which makes this more generic.
> 
> If the policy options are changed, the call to call_netevent_notifiers
> needs to be updated to handle a failure. For example, the mlxsw handler
> needs to be able to veto an option it does not support.

Did the author consider using a UDP-based tunnel like FOU? Then L4 mode
should work just fine. I believe most hardware routers do not take inner
headers into account either.

https://lwn.net/Articles/614348/
https://people.netfilter.org/pablo/netdev0.1/papers/UDP-Encapsulation-in-Linux.pdf