> I noticed that the "ib_srpt" driver does an explicit ib_modify_port in > order to set this flag. If there were a user-level version of this > function, I could do the same.
It would be pretty straightforward to add something like /dev/infiniband/isdmX that behaves like the issmX files we already have. Or we could even have something that automatically sets the IsDM bit when the first agent for DM class is created and clears it when the last agent is destroyed. (In fact we could do the same thing for IsCM if we wanted to) > I would like to propose a better interface. What if there were a generic > DM agent in the kernel that provided an API for target devices (kernel and > user) to register IOC's with it? It might look something like this: I'm not sure having a DM agent in the kernel is worth it. Why not have a generic daemon in userspace to do all the DMA stuff? I don't see a strong reason to put it in the kernel, and userspace code is quite a bit easier to write... - 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
