> Note that not that OFED 1.1 and 1.2 only include kernel drivers > which are not upstream, some of them (eg SDP, RDS) never passed any > --review-- cycle at the relevant mailing lists > (openib,netdev,lkml). Now, for OFED 1.3 there's a suggestion to add > rNFS which was also never reviewed.
Good point. I'm actually less concerned about entirely new modules than about patches to existing modules, because I think it's pretty easy for someone to understand, "oh, that module isn't in Linus's kernel yet, so if I switch to a vanilla kernel I don't have it." On the other hand, if we sneak fixes and changes into OFED that don't go upstream, then I think users and developers may waste a lot of time debugging things that someone else debugged already. With that said, perhaps it is a good idea to be stricter about getting things in the upstream kernel. For example, maybe we should make the rule that a module cannot be called "GA" for OFED if it is not merged upstream -- everything not upstream is automatically a "technology preview." This actually protects users if a module has to change when it is merged. - R. _______________________________________________ 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
