On Wed, 31 Oct 2012 07:01:49 -0700
John Fastabend [off-list ref] wrote:

On 10/29/2012 5:57 PM, Stephen Hemminger wrote:

quoted

Expose bridge port parameters over netlink. By switching
from a single byet to a nested message, this can be used for
other bridge parameters.

Although, this changes IFLA_PROTINFO attribute from one byte to a full nested
set of attributes; it is safe for applications because the
old message used IFLA_PROTINFO and new one uses
  IFLA_PROTINFO | NLA_F_NESTED.

The code still accepts to old format requests, and therefore stays
compatiable with user mode RSTP daemon. Since the type field
for nested and unnested attributes are different, and the old
code in libnetlink doesn't do the mask, it is also safe to use
with old versions of bridge monitor command.

Note: although mode is only a boolean, treating it as a
full byte since in the future someone will probably want to add more
values (like macvlan has).

Signed-off-by: Stephen Hemminger<redacted>

---

Stephen,

Did you see these two patches

http://patchwork.ozlabs.org/patch/193900/
http://patchwork.ozlabs.org/patch/193901/

here I added nested bridge attributes to IFLA_AF_SPEC and pass them down
to the drivers as needed. Should we merge these two sets so that we have
only a single nested set of bridge attributes? Either in IFLA_AF_SPEC or
IFLA_PROTINFO.

The attributes in this patch are port specifics and the one in the
patches listed above are bridge specific so in this sense perhaps
its OK to keep them separate. I'm not sure it matters much either
way but thought I would mention it.

Also I suspect these two series will have conflicts but I haven't tried
yet.

Thanks,
John

This is an area where there is no clear choice.
I would like to keep AF_UNSPEC for non-protocol stuff,
that is why I targeted PF_BRIDGE:IFLA_PROTINFO.

Other alternative would be to add sysctl which is less message
based, but is more general. (ie. /default and /all are available).