>Well the ACK for the direction switch is special, isn't it? >All I'm saying, let's pass it up to the application.
I really don't think that this is the direction that we want to take the interface. A multithreaded application could see the ACK before the request. Multiple ACKs could be received for the same request, or no ACK could be received at all. This pushes timeout handling and duplicate detection up to the any application using DS RMPP. We should work for a simpler interface, especially one exposed to userspace. Let's start with an interface that's efficient and works well in the kernel, and then determine how to expose that interface up to userspace. Let's try to keep the complexity in one location. Btw, it looks like Jack's patch has the MAD layer read MAD data while it is in transfer. I don't think that we can do this while the data is mapped. - Sean _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
