On Fri, Mar 18, 2005 at 03:25:04PM +0800, Timothy\(???x\) wrote:
> Hi James,
>=20
> Thanks for your help. I can get the latest code from the "P4 server" now.
>=20
> I get the latest code that you just update, thanks. I am going to test it.
>=20
> In your explanation, I can clearly understand the state and problem of th=
is project.
>=20
> I have few more questions to make sure my thought is right:
>=20
> 1. In "quagga-ldp", how does it implement "label" related operation (e.g.=
after label binding, write this label to the kernel "NHLFE"....)? Or in th=
is version there has no implementation of this part?
In quagga-ldp, the ldpd daemon issued label operations directly to the
kernel. In quagga-mpls, label operations are sent via the zserve API and
the zebra daemon then issues them to the kernel.
> 2. So the "quagga-mpls" will let quagga can work on mpls enhanced kernel,=
and the implementation will follow zclient & zserver communication model?=
=20
Yes.
> 3. I should patch my kernel with mpls enhancement, if I wnat to run the "=
quagga-mpls" code?
The only reason you will need to run on a mpls enabled kernel is
if you want to test the data path. I think there are still a couple of
significant bugs that will prevent us from getting that far, so at this=20
point I don't think you _need_ to run an mpls enabled kernel.
>=20
> Thanks
>=20
> Best Regards
>=20
> Timothy
> 2005.03.18
--=20
James R. Leu
jl...@mi...
|
