On Fri, Feb 4, 2011 at 10:34 AM, Ben Pfaff <[email protected]> wrote: > I had made a note to myself to take another look at netlink_set_err() > after the netlink series was put to bed. Now that I look at it again, I > think that we need to revisit how we do error reporting. > > The problem is that netlink_set_err() was not exported from 2.6.22 > through 2.6.29 and I can see no way to implement it for those versions. > If we want reliable error reporting for those versions, I think that we > will have to do careful preallocation of reply skbs in all the cases > where we currently call netlink_set_err(). > > What do you think?
I don't too much like the idea of doing this solely for compatibility reasons but it also avoids an additional point of failure, so it seems reasonable. In any case, I don't really have a better idea. It doesn't work in the upcall case but there's probably nothing we can do about that. _______________________________________________ dev mailing list [email protected] http://openvswitch.org/mailman/listinfo/dev_openvswitch.org
