Yossi Etigin wrote:
> How about marking neighbours invalid (instead of paths)
> then you could mark all neighbours invalid on flush events,
> and in ipoib_start_xmit() add neigh->valid to the gid memcmp
> condition, so that if neigh->valid is 0 ipoib will free it,
> and call ipoib_path_loookup()?
> 

Destroying the neigh destroys the ah and if we are going this way we lose 
the benefits of FLUSH_LIGHT event which is keeping the ah with the assumption 
that 
it is still valid (which is the case most of the times)

_______________________________________________
general mailing list
[email protected]
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general

To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general

Reply via email to