cvs commit: src/sys/net if_ethersubr.c

[Ruslan Ermilov]

On Tue, Feb 15, 2005 at 09:48:52AM -0800, Sam Leffler wrote:
> If the intent is simply to eliminate the dispatch of frames out of the 
> driver when there are no vlan consumers then it would seem another 
> option is to do this at the source.  I suggested this initially--that 
> VLAN_INPUT_TAG be augmented to discard frames when if_nvlans is zero 
> (this would also eliminate gratuitous tag allocation).  I believe you're 
> saying that if_nvlans can be zero and there may still be consumers for 
> the packets in which case we might look at a better way of indicating 
> there are consumers for the packets assuming we can be sure they won't 
> reach ether_demux (which might be hard).
> 
Yes, this can be hard -- consumers may live on another machine,
read: there's no direct link to the underlying hardware that
supplies VLAN frames.

> As to your other suggestion of allocating an mbuf flag bit that'd be 
> fine with me.  I didn't offer that because I thought we were out of free 
> bits but I see m_flags got changed to an int a while back and only 
> 16-bits are assigned.
> 
OK, I'll post the patches soon then.


Cheers,
-- 
Ruslan Ermilov
ru at FreeBSD.org
FreeBSD committer
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/cvs-src/attachments/20050215/dd20c651/attachment.bin