Re: switchdev and VLAN ranges

[Vivien Didelot <vivien.didelot@savoirfairelinux.com>]

On Oct. Sunday 11 (41) 09:12 AM, Jiri Pirko wrote:
> Sat, Oct 10, 2015 at 12:36:26PM CEST, nikolay@cumulusnetworks.com wrote:
> >On 10/10/2015 09:49 AM, Elad Raz wrote:
> >> 
> >>> On Oct 10, 2015, at 2:30 AM, Vivien Didelot <vivien.didelot@savoirfairelinux.com> wrote:
> >>>
> >>> I have two concerns in mind:
> >>>
> >>> a) if we imagine that drivers like Rocker allocate memory in the prepare
> >>> phase for each VID, preparing a range like 100-4000 would definitely not
> >>> be recommended.
> >>>
> >>> b) imagine that you have two Linux bridges on a switch, one using the
> >>> hardware VLAN 100. If you request the VLAN range 99-101 for the other
> >>> bridge members, it is not possible for the driver to say "I can
> >>> accelerate VLAN 99 and 101, but not 100". It must return OPNOTSUPP for
> >>> the whole range.
> >> 
> >> Another concern I have with vid_being..vid_end range is the “flags”. Where flags can be BRIDGE_VLAN_INFO_PVID.
> >> There is no sense having more than one VLAN as a PVID.
> >> This leave the HW vendor the choice which VLAN id they will use as the PVID.
> >> 
> >
> >iproute2 doesn't allow to do it but I can see that someone can actually make it
> >so the flags for the range have it and it doesn't look correct. Perhaps we need
> >something like the patch below to enforce this from kernel-side.
> >
> >
> >diff --git a/net/bridge/br_netlink.c b/net/bridge/br_netlink.c
> >index d78b4429505a..02b17b53e9a6 100644
> >--- a/net/bridge/br_netlink.c
> >+++ b/net/bridge/br_netlink.c
> >@@ -524,6 +524,9 @@ static int br_afspec(struct net_bridge *br,
> > 			if (vinfo_start)
> > 				return -EINVAL;
> > 			vinfo_start = vinfo;
> >+			/* don't allow range of pvids */
> >+			if (vinfo_start->flags & BRIDGE_VLAN_INFO_PVID)
> >+				return -EINVAL;
> > 			continue;
> > 		}
> > 
> 
> Looks correct to me. Could you please submit this properly? Thanks!

The above patch is correct, but we only solve part of the problem, since
the range and bridge flags are exposed by switchdev_obj_port_vlan as is.

Thanks,
-v